Hybrid adaptive control for the DC-DC boost converter

[0001] This application claims the benefit under 35 U.S.C. Section 119(e) of the following and commonly-assigned U.S. Provisional Application Ser. No. 63/354,466, filed on Jun. 22, 2022, by Ryan S. Johnson, Berk Altin, and Roberto G. Sanfelice, entitled “HYBRID ADAPTIVE CONTROL FOR THE DC-DC BOOST CONVERTER,”, which application is incorporated by reference herein.

[0002] This invention was made with Government support under Grant nos. ECS-1710621 and CNS-1544396 awarded by the National Science Foundation (NSF) and Grant nos FA9550-16-1-0015, FA9550-19-1-0053, and FA9550-19-1-0169 awarded by Airforce Office of Scientific Research (AFOSR). The Government has certain rights in the invention.

[0003] The present invention relates to DC-DC boost converter control.

[0004] The DC-DC boost converter is widely used in the power systems of electric vehicles [1]. These systems operate under constantly changing demands such as supplying energy during acceleration and storing it during braking, necessitating power conversion technology that is capable of adapting to these changes [2]. The industry standard control scheme for the boost converter is pulse-width modulation (PWM). However, since PWM controllers typically utilize a linearized model of the converter dynamics, the stability properties only hold locally near the set-point [3]. Recently, a renewed interest in power converters has originated from the rise of hybrid modeling paradigms [4-11] and new perspectives on their control, including time-based switching, state event triggered control, and optimization-based control, have been proposed. However, improved methods of stabilizing boost converters under parameter uncertainty are needed. The present disclosure satisfies this need.

[0005] A system comprising and implementing an adaptive control algorithm for a DC-DC boost converter, the algorithm estimating the input voltage and output load of the converter in finite time. Using these estimates, a control algorithm that “unites” global and local control schemes. The global control scheme induces practical asymptotic stability of a desired output voltage and corresponding current, and the local control scheme can maintains industry-standard Pulse Width Modulation (PWM) behavior during steady state. The control algorithm can be implemented with one or more computers or circuits.

[0006] Illustrative embodiments of the present invention include, but are not limited to, the following examples.

[0007] 1. A boost converter, comprising:

• • [0008] a boost converter circuit comprising a switch comprising a switch state (open or closed), a capacitor, an inductor, and a diode;
  • [0009] a hybrid adaptive controller connected to the switch for applying a hybrid algorithm controlling the switch state when the hybrid algorithm is executed by the hybrid adaptive controller; wherein the hybrid adaptive controller further comprises:
  • [0010] a global controller:
  • [0011] determining a first switch state command (open or close) as a function of estimates of the input voltage and a load resistance received from a parameter estimator, a measurement of an output voltage across the capacitor and/or a measurement of the inductor current passing through the inductor, and
  • [0012] outputting the first switch state command to the switch to set the switch state causing the output voltage to converge to a desired output voltage, thereby raising an input voltage E, inputted to the boost converter circuit, to the desired output voltage across the capacitor when a load comprising the load resistance is connected across the capacitor;
  • [0013] a local controller:
  • [0014] determining a second switch state command as a function of the estimates of the input voltage and the load resistance received from the parameter estimator, and the measurement of the output voltage and/or the measurement of the inductor current, and
  • [0015] outputting the second switch state command to the switch to set the switch state maintaining the output voltage at or within a set range of the desired output voltage and maintaining the inductor current at or within a set range of a desired current, when the load is connected; and
  • [0016] a supervisor determining whether to execute the global controller outputting the first switch state command, or the local controller outputting the second switch state command, depending on the measurement of the output voltage and the measurement of the inductor current.

[0017] 2. The boost converter of example 1, wherein the hybrid adaptive controller further comprises the parameter estimator.

[0018] 3. The boost converter of example 1, wherein the supervisor uses the output voltage and the inductor current as feedback to select:

• • [0019] the global controller outputting the switch state command until the output voltage has converged to within a first voltage range of the desired output voltage and the inductor current has converged to within a first current range of a desired current, or
  • [0020] the local controller to output the second switch state command so long as the desired output voltage is maintained within a second voltage range and the inductor current is maintained within a second current range.

[0021] 4. The boost converter of example 1, wherein:

• • [0022] the global controller induces asymptotic stability of the output voltage at the desired output voltage and the inductor current at the desired current, and
  • [0023] the local controller maintains, in a the output voltage at the desired output voltage and the inductor current at the desired current.

[0024] 5. The boost converter of example 1, wherein the hybrid adaptive controller comprises one or more processors and one or more applications or programs executed by the one or more processors execute the hybrid algorithm comprising:

• • [0025] a parameter estimating algorithm, expressed using an estimator hybrid system, outputting the estimated load resistance and estimated input voltage in response to a parameter input comprising an initial input voltage;
  • [0026] a global control algorithm, expressed as a global controller hybrid system; outputting the first switch state command as a function of the output voltage and the inductor current,
  • [0027] a local control algorithm, expressed as a local controller hybrid system, outputting the second switch state command as a function of the output voltage and the inductor current; and
  • [0028] a supervisor algorithm, expressed as a supervisor hybrid system, outputting:
  • [0029] a first state instructing the hybrid adaptive controller to output the first switch state command to the switch until the output voltage is within a first voltage range and the inductor current is within a second current range, at which point outputting a second state, or
  • [0030] a second state instructing the hybrid controller to output the second switch state command so long as the output voltage is within a second voltage range and the inductor current is within a second current range.

[0031] 6. The boost converter of example 5, wherein the first state and the second state comprise logic states 0 or 1.

[0032] 7. The boost converter of example 5, wherein:

• • [0033] the parameter estimating algorithm is initialized with at least one of the current or the voltage set to zero, so that the first set of values N have converged to the desired output voltage and/or desired current, for any value and polarity of the desired output voltage or the desired current.

[0034] 8. The boost converter of example 5, wherein the one or more applications or programs obtain or create a representation of the hybrid systems and determine the switch state commands by processing, in the hybrid systems, state variables representing the switch state commands as a function of the output voltage and/or desired current.

[0035] 9. The boost converter of example 8, wherein the processing of the state variables comprises modifying the state variables via discrete jumps and continuous evolution in the hybrid systems.

[0036] 10. The boost converter of example 8, wherein the hybrid systems determine the switch state commands from the estimates of the load resistance and the input voltage, and the measurements of the output voltage and the current, using ohm's law relationships between the inductor current, the output voltage, the input voltage, and the load resistance.

[0037] 11. The boost converter of example 3, wherein:

• • [0038] the first voltage range and the first current range each comprise a first set of values N to which maximal solutions of the output voltage and the inductor current, respectively, converge in finite time in response to the first switch state command, and
  • [0039] the second voltage range and the second current range each comprise a second set of values M to which the output voltage and the inductor current, respectively, remain in steady state in response to the second switch state command.

[0040] 12. The boost converter of example 11, wherein the first set of values N comprises positive values of the output voltage across the capacitor and/or positive values of a current passing through the inductor.

[0041] 13. The boost converter of example 1, wherein the local controller hybrid system implements a timer.

[0042] 14. The boost converter of example 1, wherein at least one of the global controller, local controller, or the supervisor comprise one or more circuits comprising one or more application specific integrated circuits or one or more field programmable gate arrays.

[0043] 15. The boost converter of example 1, wherein the parameter estimator is expressed using an estimator hybrid system.

[0044] 16. A computer implemented method for controlling a boost converter, comprising:

• • [0045] obtaining a boost converter circuit comprising a switch comprising a switch state (open or closed), a capacitor, an inductor, and a diode;
  • [0046] applying a hybrid algorithm controlling the switch state, comprising:
  • [0047] using a global controller:
  • [0048] determining a first switch state command (open or close) as a function of estimates of the input voltage and a load resistance received from a parameter estimator, a measurement of an output voltage across the capacitor, and a measurement of the inductor current passing through the inductor, and
  • [0049] outputting the first switch state command to the switch to set the switch state causing the output voltage to converge to a desired output voltage, thereby raising an input voltage E, inputted to the boost converter circuit, to the desired output voltage across the capacitor when a load comprising the load resistance is connected across the capacitor;
  • [0050] using a local controller:
  • [0051] determining a second switch state command as a function of the estimates of the input voltage and the load resistance received from the parameter estimator, and the measurement of the output voltage and the measurement of the inductor current, and
  • [0052] outputting the second switch state command to the switch to set the switch state maintaining the output voltage at or within a set range of the desired output voltage and maintaining the inductor current at or within a set range of a desired current, when the load is connected; and
  • [0053] determining, in a supervisor, whether to execute the global controller outputting the first switch state command, or the local controller outputting the second switch state command, depending on the measurement of the output voltage and the measurement of the inductor current.

[0054] 17. A computer implemented system for controlling a boost converter circuit, comprising:

• • [0055] one or more processors, or a computer readable medium configured for:
  • [0056] receiving at least one of an desired output voltage or an inductor current for application to an output load using the boost converter circuit comprising a switch comprising a switch state (open or closed), a capacitor, an inductor, and a diode;
  • [0057] determining a first switch state command (open or closed) as a function of estimates of an input voltage and a load resistance received from a parameter estimator, a measurement of an output voltage across the capacitor, and a measurement of the inductor current passing through the inductor, wherein the first switch state command sets the switch state causing the output voltage to converge to a desired output voltage, thereby raising an input voltage E, inputted to the boost converter circuit, to the desired output voltage across the capacitor when a load comprising the load resistance is connected across the capacitor; and
  • [0058] determining a second switch state command as a function of the estimates of the input voltage and the load resistance received from the parameter estimator, and the measurement of the output voltage and the measurement of the inductor current, wherein the second switch state command sets the switch state maintaining the output voltage at or within a set range of the desired output voltage and/or maintains the inductor current at or within a set range of a desired current, when the load is connected; and
  • [0059] determining whether to output the first switch state command or the second switch state command to control the switch, depending on the measurement of the output voltage and the measurement of the inductor current.

[0060] 18. The system of example 17 comprising the one or more processors and one or more applications or programs executed by the one or more processors for executing a hybrid algorithm performing the determining steps and outputting the switch state commands.

[0061] 19. The system of example 17, wherein at least one of the global controller, local controller, or the supervisor comprise one or more circuits comprising one or more application specific integrated circuits or one or more field programmable gate arrays.

[0073] In the following description of the preferred embodiment, reference is made to the accompanying drawings which form a part hereof, and in which is shown by way of illustration a specific embodiment in which the invention may be practiced. It is to be understood that other embodiments may be utilized and structural changes may be made without departing from the scope of the present invention.

[0074] Embodiments of the present invention provide a system for controlling pulse width modulation of a DC-DC boost converter circuit using a control algorithm that induces practical asymptotic stability of a desired set point for the DC-DC boost converter even under uncertainty in the circuit input voltage and output load resistance. Further, the control algorithm maintains PWM behavior during steady-state operation.

[0075] FIG. 1 illustrates an example boost converter circuit, comprising an inductor, a switch, a diode, and a capacitor, coupled to a hybrid adaptive controller controlling the output voltage via pulse width modulation of the switch. FIG. 2 illustrates an example hybrid adaptive controller 200 comprising a parameter estimator 202; a converter estimator 202; a supervisor 204; a local controller 206; and a global controller 208. The parameter estimator estimates the input voltage and load resistance which are used by the converter estimator to estimate the output voltage. Depending on the output voltage, the supervisor selects the local controller or the global controller to control the position of the switch (open or closed) in the boost converter circuit.

[0076] The parameter estimator, converter estimator, global controller, local controller, and supervisor execute algorithms represented by hybrid systems, as discussed in the following sections.

[0077] An example hybrid system has data (C, F, D, G, κ) and is given by [14,15]

[0078] $\mathscr{H}=\{\begin{array}{cc}\left(x,u\right)\in C& \stackrel{.}{x}\in F\left(x,u\right)\\ \left(x,u\right)\in D& x^{+}\in G\left(x,u\right)\\ y& =\kappa \left(x,u\right)\end{array}$

[0079] where x∈ⁿ is the state, u∈^m is the input, F:ⁿ×^mn is a set-valued map defining the flow map of the differential inclusion capturing the continuous dynamics, and C⊂ⁿ×^m defines the flow set on which flows are permitted. Similarly, G:ⁿ×^mn is a set-valued map defining the jump map of the difference inclusion modeling the discrete behavior, and D⊂ⁿ×^m is the jump set on which jumps are permitted. The vector y∈ⁿ defines the output of the hybrid system.

[0080] FIG. 1 illustrates a boost converter comprising is a class of switched-mode power supply that utilizes a switch S, inductor L, diode d, and capacitor c to raise the voltage at the output load R compared to the input voltage E. The state of the switch S (open or closed) represents the control input to the boost converter plant. When the switch is closed, current flows through the inductor and generates a magnetic field. When the switch is opened, the inductor magnetic field decays to maintain the current towards the load, causing a polarity reversal within the inductor. The primary voltage source in series with the inductor then charges the capacitor through the diode to a higher voltage than is attainable using the voltage source alone. If the switch is cycled fast enough, the inductor does not fully discharge between cycles and the load voltage remains higher than that of the source.

[0081] The boost converter dynamics may be expressed as a (continuous time) plant, _P, with discrete-valued input denoting the position of the switch S. We model it as in (1) but with no jumps. That is, _P with state

[0082] $𝓍:=\left(v_c,i_L\right),x\in {𝒳}_P:={\tilde{M}}_0\bigcup {\tilde{M}}_1$
which are defined below, input q∈{0, 1}, and output x. Following [12], its dynamics reduce to the differential inclusion with constraints

[0083] $\stackrel{.}{x}\in F_P\left(x,q\right)\left(x,q\right)\in C_P$ $\mathrm{where}{C}_P:=\left({\tilde{M}}_0\times \left\{0\right\}\right)\bigcup \left({\tilde{M}}_1\times \left\{1\right\}\right)$ ${\tilde{M}}_0:=\stackrel{\_}{M_1}\bigcup \stackrel{\_}{M_3}=\left\{x\in {ℝ}^2:i_L\ge 0\right\},$ ${\tilde{M}}_1:=M_2=(x\in {ℝ}^2:v_c\ge 0\},\mathrm{with}$ $M_1=\left\{x\in {ℝ}^2:i_L>0\right\}\bigcup \left\{x\in {ℝ}^2:v_c\le E,i_L=0\right\}$ $M_2=\left\{x\in {ℝ}^2:v_c\ge 0\right\}$ $M_3=\left\{x\in {ℝ}^2:v_c>E,i_L=0\right\}$

• • [0084] and F_P is the Krasovskii regularization of the vector fields and the corresponding constraint sets associated with each mode of circuit operation given by, for each x∈χ_P, see [12] for details.

[0085] $F_P\left(x,0\right):=\{\begin{array}{c}\left[\begin{array}{c}-\frac{1}{\mathrm{Rc}}{v}_c+\frac{1}{c}{i}_L\\ -\frac{1}{L}{v}_c+\frac{E}{L}\end{array}\right]\mathrm{if}x\in \stackrel{\_}{M_1}\backslash \stackrel{\_}{M_3}\\ \left\{-\frac{1}{\mathrm{Rc}}{v}_c\right\}\times \left[-\frac{1}{L}{v}_c+\frac{E}{L},0\right]\mathrm{if}x\in \stackrel{\_}{M_3}\end{array}{F}_P\left(x,1\right):=\left[\begin{array}{c}-\frac{1}{\mathrm{Rc}}{v}_c\\ \frac{E}{L}\end{array}\right]\mathrm{if}x\in {\tilde{M}}_1.$

[0086] Since the supply voltage E and load resistance R for the converter may vary during operation, it is desirable to estimate these values so the chosen control algorithm can adapt accordingly. Furthermore, since the stability properties of boost converter PWM control algorithms typically only hold locally near the set-point, we design an algorithm that permits on-line estimation of the boost converter input voltage E and load resistance R and an adaptive control law whose closed-loop system induces practical asymptotic stability of a desired output voltage.

[0087] The framework combines two feedback controllers and a logic-based algorithm that selects which controller to apply. Uniting control strategies permit combining a global controller that renders a set-point stable but may not have good performance near the set point, and a local controller that induces satisfactory performance, but only locally [13].

[0088] The uniting control logic is implemented as follows. Given a plant _P as in (2) interconnected with two separate control algorithms, ₀, referred to as “global,” and ₁, referred to as “local,” the choice of control algorithm is governed by a supervisory controller that selects between ₁ and ₀ based on the plant state in relation to a closed set and an open set ⊃ as follows:—Apply the global controller ₁ until the solution to the plant enters . When any such point is reached, switch to the local controller ₀ and apply the local controller ₀ as long as the solution to the plant remains inside . If the state of the plant leaves , switch to the global controller ₁.

[0089] Given ₁ and ₀, the uniting control sets and can be designed to satisfy the assumption that each maximal solution x to _P with input q generated by ₁ converges to in finite (hybrid) time; and each solution x to _P from and input q generated by ₀ remains in for all (hybrid) time.

[0090] Hence, for the boost converter, the global controller drives the converter state into , from where solutions under the local PWM controller remain inside for all future time.

[0091] Next we address Problem 1 from Example 3. Suppose that the unknown converter parameters R and E take values from R∈[R_min, R_max] and E∈[E_min, E_max] where R_min, R_max, E_min, E_max>0 are known. Then, for the purpose of estimating R and E, we establish the following lemma that allows us to express the dynamics of maximal solutions to _P in a convenient form.

[0092] Lemma 4.1. Each maximal solution tx(t) to _P, in (2) with input tq(t) satisfying
{dot over (x)}(t)=f₁(x(t),q(t))+f₂(x(t),q(t))ϑ  (4)

[0093] Since _P in (2) is a continuous-time system, its solutions are parameterized using only t. for all t∈dom(x, q), where ϑ=(ϑ₁, ϑ₂):=(R⁻¹, E) and

[0094] $f_1\left(x,q\right):=\{\begin{array}{ccc}\left[\begin{array}{c}\frac{{}^{i}L}{c}\\ -\frac{{\upsilon }_c}{L}\end{array}\right]& \mathrm{if}& q=0,x\in \stackrel{\_}{M_1}\backslash \stackrel{\_}{M_3}\\ \left[\begin{array}{c}0\\ 0\end{array}\right]& \mathrm{if}& \left(q=1,x\in \widetilde{M_1}\right)\mathrm{or}\left(q=0,x\in \stackrel{\_}{M_3}\right)\end{array}{f}_2\left(x,q\right):=\{\begin{array}{ccc}\left[\begin{array}{cc}-\frac{{\upsilon }_c}{c}& 0\\ 0& \frac{1}{L}\end{array}\right]& \mathrm{if}& \begin{array}{c}\left(q=0,x\in \stackrel{\_}{M_1}\backslash \stackrel{\_}{M_3}\right)\\ \mathrm{or}\left(q=1,x\in \widetilde{M_1}\right)\end{array}\\ \left[\begin{array}{cc}-\frac{{\upsilon }_c}{c}& 0\\ 0& 0\end{array}\right]& \mathrm{if}& q=0,x\in \stackrel{\_}{M_3}\end{array}.$

[0095] Estimating the parameters R and E is equivalent to estimating the parameter vector ϑ in (4). For this purpose, the finite-time parameter estimator in [16,17] is extended to classes of hybrid systems whose solutions satisfy (4). The algorithm is expressed as a hybrid system, denoted _E, and operates as follows. Let z_E=({circumflex over (x)}, {circumflex over (θ)}, {circumflex over (ω)}, Q, η, Γ) be a solution to _E—hence, defined on a hybrid time domain—where {circumflex over (x)} is the estimate of x, {circumflex over (ϑ)} is the estimate of ϑ, and ω, Q, η, Γ are auxiliary variables. Consider the initial interval of flow I⁰:={t:(t, 0)∈dom z_E} in dom z_E with constant ϑ and initial conditions ω(0,0)=0,Q(0,0)=0,η(0,0)=0,Γ(0,0)=0, and ϑ(0,0)∈² arbitrary. Omitting the (t,j) of solutions, for the sake of making an argument, suppose that over this interval of flow, Q and Γ satisfy
{dot over (Q)}=ω^τω,{dot over (Γ)}=ω^τωϑ
Then, if there exists a positive time t₁∈I⁰ such that Q(t₁, 0) is invertible, resetting {circumflex over (ϑ)} to the value of Q⁻¹Γ leads to

[0096] $\begin{array}{cc}\hat{\vartheta }\left(t_1,1\right)=Q^{-1}\left(t_1,0\right)\Gamma \left(t_1,0\right)& \left(5\right)\end{array}$ $\begin{array}{cc}={\left({\int }_0^{t_1}{\omega \left(t,0\right)}^{\top }\omega \left(t,0\right)\mathrm{dt}\right)}^{-1}\left({\int }_0^{t_1}{\omega \left(t,0\right)}^{\top }\omega \left(t,0\right)\mathrm{dt}\vartheta \right)=\vartheta & \left(6\right)\end{array}$

[0097] However, since ϑ is unknown prior to hybrid time (t₁, 1), a trajectory for Γ satisfying (5) cannot be generated. Due to this, we rewrite the dynamics of Γ as

[0098] $\begin{array}{c}\stackrel{.}{\Gamma }={\omega }^{\top }\left(\omega \vartheta +\omega \left(\hat{\vartheta }-\hat{\vartheta }\right)+\left(x-x\right)+\left(\hat{x}-\hat{x}\right)\right)\\ ={\omega }^{\top }\left(\omega \hat{\vartheta }+x-\hat{x}-\eta \right)\end{array}$

[0099] where η=x−{circumflex over (x)}−ω(ϑ−{circumflex over (ϑ)}). Note that since ω(0,0)=0, the initial condition η(0,0)=0 implies {circumflex over (x)}(0,0)=x(0,0). Differentiating η yields
{dot over (n)}={dot over (x)}−{circumflex over ({dot over (x)})}−{dot over (ω)}(ϑ−{circumflex over (ϑ)})+ω{circumflex over ({dot over (ϑ)})}

[0100] Next we define a matrix function (x, q)K(x, q)=K^τ(x,q)>0 that is to be designed. The arguments of K are omitted below for simplicity. Let {circumflex over (x)}, ω, and {circumflex over (ϑ)} satisfy
{circumflex over ({dot over (x)})}=f₁(x,q)+f₂(x,q){circumflex over (ϑ)}+K(x−{circumflex over (x)})+ω{circumflex over ({dot over (ϑ)})}
{dot over (ω)}=f₂(x,q)−Kω.

[0101] Plugging the expressions above into (7) yields

[0102] $\begin{array}{cc}\begin{array}{c}\stackrel{.}{\eta }=f_2\left(x,q\right)\left(\vartheta -\hat{\vartheta }\right)-K\left(x-\hat{x}\right)-\omega \hat{\vartheta }-\left(f_2\left(x,q\right)-K\omega \right)\left(\vartheta -\hat{\vartheta }\right)+\omega \stackrel{.}{\hat{\vartheta }}\\ =-K\left(x-\hat{x}\right)+K\omega \left(\vartheta -\hat{\vartheta }\right)=-K\eta \end{array}& \left(7\right)\end{array}$

[0103] Hence, ω, Q, η, and Γ are now expressed in terms of known quantities and we can compute {circumflex over (ϑ)} as in (6).

[0104] Following [17], the estimation scheme outlined above can be implemented as a hybrid algorithm whose jump map imposes the initial conditions specified above (5) and computes {circumflex over (ϑ)} as in (6). The hybrid system, denoted by

[0105] ${\mathscr{H}}_E=\left(C_E,F_E,D_E,G_E,\hat{\theta }\right),\mathrm{has}\mathrm{state}$ $z_E:=\left(\hat{x},\hat{\vartheta },\omega ,Q,\eta ,\Gamma \right)\in {𝒳}_E:={ℝ}^2\times {ℝ}^2\times {ℝ}^{2\times 2}\times {\mathrm{𝒮𝒫}}^{2\times 2}\times {ℝ}^2\times {ℝ}^2,$ $\mathrm{inputs}\left(x,q\right)\in C_P,\mathrm{output}\hat{\theta }\in {ℝ}_{>0}^2,\mathrm{and}\mathrm{dynamics}$
and dynamics

[0106] $\begin{array}{cc}\begin{array}{cc}{\stackrel{.}{z}}_E=F_E\left(x,q,z_E\right)& \left(x,q,z_E\right)\in C_E\\ z_E^{+}=G_E\left(x,z_E\right)& \left(x,q,z_E\right)\in D_E\\ \hat{\theta }=h_E\left(\hat{\vartheta }\right)& \end{array}& \left(8\right)\end{array}$

• • [0107] where G_E(x,z_E):=(x, Q⁻¹Γ, 0,0,0,0),

[0108] $F_E\left(x,q,z_E\right):=\left[\begin{array}{c}{f}_1\left(x,q\right)+f_2\left(x,q\right)\hat{\vartheta }+K\left(x-\hat{x}\right)+\omega h\left(x,q,z_E\right)\\ h\left(x,q,z_E\right)\\ f_2\left(x,q\right)-K\omega \\ {\omega }^{\top }\omega \\ -K\eta \\ {\omega }^{\top }\left(\omega \hat{\vartheta }+x-\hat{x}-\eta \right)\end{array}\right]$

[0109] The matrix function K and the parameter Ω=Ω^τ>0 modify the convergence rate of {circumflex over (x)} and {circumflex over (ϑ)} during flows, and μ>0 ensures that Q⁻¹ is well-defined in the jump map. The dynamics of _E in (8) are similar to the estimator proposed in [17]. However, in [17], f₁ and f₂ are continuous functions of the state and input, compared to piecewise continuous functions in (4).

[0110] To ensure completeness of maximal solutions for the control algorithms discussed in the following subsections, we require that

[0111] $\left(\hat{R},\hat{E}\right)\in {ℝ}_{>0}^2.$
Hence, the output of _E is computed as

[0112] $h_E\left(\hat{\vartheta }\right):=\left[\begin{array}{c}{\varrho \left({\hat{R}}^{-1},R_{\max }^{-1},R_{\min }^{-1}\right)}^{-1}\\ \varrho \left(\hat{E},E_{\min },E_{\max }\right)\end{array}\right]\mathrm{where}\varrho \left(s,s_{\min },s_{\max }\right):=\{\begin{array}{ccc}{s}_{\min }& \mathrm{if}& s\le s_{\min }\\ s& \mathrm{if}& s_{\min }<s<s_{\max }\\ s_{\max }& \mathrm{if}& s_{\max }\le s\end{array}.$

[0113] Similarly to [16,17], each maximal solution to _E is guaranteed to jump for certain assumptions (see Appendix).

[0114] The hybrid control algorithm in [12] represents an ideal candidate for the global controller. Given a desired output voltage v_c*, this algorithm renders the set

[0115] ${𝒜}_P:=\left\{x\in {ℝ}^2:v_c=v_c^{*},i_L=i_L^{*}=\frac{v_c^{*2}}{\mathrm{RE}}\right\}$

• • [0116] globally asymptotically stable for the boost converter when the converter parameters c, L, R, E>0 are known.

[0117] However, in contrast to [12], the parameters R and E are unknown in this example. Hence, we employ the certainty equivalence principle and substitute the parameter estimates {circumflex over (R)} and Ê from _E in (8) for R and E, respectively. Then, following the derivation in [12], given a desired voltage

[0118] $v_c^{*},$
the setpoint

[0119] ${\hat{\iota }}_L^{*}:=\frac{v_c^{*2}}{\hat{R}\hat{E}}$
is stabilized using the control Lyapunov function

[0120] $V\left(x,\hat{\theta }\right)={\left(x-x^{*}\left(\hat{\theta }\right)\right)}^{\top }P\left(x-x^{*}\left(\theta \right)\right),$
where

[0121] $P=\left[\begin{array}{cc}{p}_{11}& 0\\ 0& p_{22}\end{array}\right]>0$
with

[0122] $\frac{p_{11}}{c}=\frac{p_{22}}{L}.$
Since the interconnection of _P and _E is a hybrid system, the input and state of _P are now parameterized by (t,j). We define a hybrid system ₁ with state

[0123] $z_1:=q\in {𝒳}_1:=\left\{0,1\right\},\mathrm{inputs}x\in {𝒳}_P\mathrm{and}\hat{\theta }\in {ℝ}_{>0}^2,$
and dynamics

[0124] $\begin{array}{cc}\stackrel{.}{q}=0=:F_1\left(z_1\right)\left(x,z_1,\hat{\theta }\right)\in C_1& \left(12\right)\end{array}$ $q^{+}=1-q=:G_1\left(z_1\right)\left(x,z_1,\hat{\theta }\right)\in D_1$ ${\kappa }_1\left(x,z_1,\hat{\theta }\right):=q$

• • [0125] where κ₁ represents the input q of _P

[0126] $C_1:=\left\{\left(x,z_1,\hat{\theta }\right)\in {𝒳}_P\times {𝒳}_1\times {ℝ}_{>0}^2:{\tilde{\gamma }}_0\left(x,\hat{\theta }\right)\le \rho ,q=0\right\}$ $\bigcup \left\{\left(x,z_1,\hat{\theta }\right)\in {𝒳}_P\times {𝒳}_1\times {ℝ}_{>0}^2:{\tilde{\gamma }}_1\left(x,\hat{\theta }\right)\le \rho ,q=1\right\}$ $D_1:=\left\{\left(x,z_1,\hat{\theta }\right)\in {𝒳}_P\times {𝒳}_1\times {ℝ}_{>0}^2:{\tilde{\gamma }}_0\left(x,\hat{\theta }\right)\ge \rho ,q=0\right\}$ $\bigcup \left\{\left(x,z_1,\hat{\theta }\right)\in {𝒳}_P\times {𝒳}_1\times {ℝ}_{>0}^2:{\tilde{\gamma }}_1\left(x,\hat{\theta }\right)\ge \rho ,q=1\right\},$

• • [0127] and ρ∈_>0 is a design parameter that, as in [12], spatially regularizes the closed-loop global controller by modifying the separation between the functions {tilde over (γ)}₀ and {tilde over (γ)}₁ to avoid Zeno behavior. The functions {tilde over (γ)}₀ and {tilde over (γ)}₁ are

[0128] $\begin{array}{cc}{a}_0=-\frac{p_{11}}{\hat{R}c}& a_1=-\frac{p_{11}}{\hat{R}c}\\ b_0=\frac{p_{11}{v}_c^{*}}{\hat{R}c}+\frac{p_{22}{\hat{\iota }}_L^{*}}{L}& b_1=\frac{p_{11}{v}_c^{*}}{\hat{R}c}\\ c_0=-\frac{p_{11}{v}_c^{*}}{c}+\frac{p_{22}\hat{E}}{L}& c_1=\frac{p_{22}\hat{E}}{L}\\ d_0=-\frac{p_{22}{\hat{\iota }}_L^{*}\hat{E}}{L}& d_1=-\frac{p_{22}{\hat{\iota }}_L^{*}\hat{E}}{L}\end{array}$

• • [0129] and

[0130] $K_0=k_0\frac{2p_{11}}{\hat{R}c},K_1=k_1\frac{2p_{11}}{\hat{R}c},$
where k₀, k₁∈(0,1) are design parameters that ensure K₀, K₁∈(0,2p₁₁/({circumflex over (R)}c)).

[0131] Given (t,j)x(t,j) and (t,j){circumflex over (θ)}(t,j), each solution q to ₁ maintains a constant switch state until x(t,j) intersects with the ρ level-set of {tilde over (γ)}_q, at which point the value of q is toggled.

[0132] Note that the jump set D₁ below (12) has been modified compared to the model in [12]. In particular, the conditions {tilde over (γ)}₀(x, {circumflex over (θ)})=ρ and {tilde over (γ)}₁(x, {circumflex over (θ)})=ρ in (12) are instead {tilde over (γ)}₀(x, {circumflex over (θ)})≥ρ and {tilde over (γ)}₁(x, {circumflex over (θ)})≥ρ, respectively. This change ensures completeness of maximal solutions for the closed-loop uniting control algorithm discussed in Example 4.4.

[0133] Appendix D establishes the maximal solutions form a complete set and Appendix E establishes convergence.

[0134] In this example, the local control algorithm for the uniting control framework is described. Recall from Example 3 that we desire to maintain PWM behavior near the setpoint. Assuming the converter operates only in the continuous conduction mode, we design the PWM controller by averaging the converter dynamics as in [18]. The average system for the steady-state converter is

[0135] $\dot{x}=A_0\left(\hat{\theta }\right)x+B_0\left(\hat{\theta }\right)+d\left(x,\hat{\theta }\right)\left(A_1\left(\hat{\theta }\right)-A_0\left(\hat{\theta }\right)\right)x$

• • [0136] where d represents the PWM duty cycle, which is to be designed, and

[0137] $A_0\left(\hat{\theta }\right):=\left[\begin{array}{cc}-\frac{1}{\widehat{\mathrm{Rc}}}& \frac{1}{c}\\ -\frac{1}{L}& 0\end{array}\right],A_1\left(\hat{\theta }\right):=\left[\begin{array}{cc}-\frac{1}{\widehat{\mathrm{Rc}}}& 0\\ 0& 0\end{array}\right],B_0\left(\hat{\theta }\right):=\left[\begin{array}{c}0\\ \frac{\hat{E}}{L}\end{array}\right]$

[0138] Next, we linearize (16) about x*({circumflex over (θ)}) (see [3, 19] for details) and denote the region of the state-space where the linearization holds as ⊂². Expressing the linearized average model in error coordinates yields

[0139] $\begin{array}{cc}\stackrel{.}{\stackrel{~}{x}}=A_{\mathrm{avg}}\left(\hat{\theta }\right)x+B_{\mathrm{avg}}\left(\hat{\theta }\right)\tilde{d}\left(x,\hat{\theta }\right)\mathrm{where}& \left(17\right)\end{array}$ $\tilde{x}:=x-x^{*}\left(\hat{\theta }\right)\mathrm{and}\tilde{d}\left(x,\hat{\theta }\right):=d\left(x,\hat{\theta }\right)-d^{*}\left(\hat{\theta }\right),\mathrm{with}$ $d^{*}\left(\hat{\theta }\right):=1-\hat{E}/v_c^{*}$

• • [0140] where {tilde over (x)}:=x−x*({circumflex over (θ)}) and {tilde over (d)}(x, {circumflex over (θ)}):=d(x, {circumflex over (θ)})−d*({circumflex over (θ)}), with d*({circumflex over (θ)}):=1−Ê/v*_c being the steady-state duty cycle for the linearized average model, and

[0141] $A_{\mathrm{avg}}\left(\hat{\theta }\right):=\left[\begin{array}{cc}-\frac{1}{\hat{R}c}& \frac{\hat{E}}{v_c^{*}c}\\ -\frac{\hat{E}}{v_c^{*}L}& 0\end{array}\right],B_{\mathrm{avg}}\left(\hat{\theta }\right):=\left[\begin{array}{c}\frac{v_c^{*2}}{\hat{R}\hat{E}c}\\ \frac{v_c^{*}}{L}\end{array}\right]$

[0142] Since the pair (A_avg, B_avg) is controllable for all {circumflex over (R)}, Ê>0, we apply a full state-feedback controller of the form
{tilde over (d)}(x,{circumflex over (θ)})=−{tilde over (K)}({circumflex over (θ)}){tilde over (x)}

• • [0143] yielding the closed-loop dynamics
    {circumflex over ({dot over (x)})}=A_cl({circumflex over (θ)}){tilde over (x)}
  • [0144] where, for each

[0145] $\hat{\theta }\in {ℝ}_{>0}^2,$
{tilde over (K)} is chosen so that the matrix A_cl({circumflex over (θ)}):=A_avg({circumflex over (θ)})−B_avg ({circumflex over (θ)}){tilde over (K)}({circumflex over (θ)}) is Hurwitz. Then, the PWM duty cycle is computed as
d(x,{circumflex over (θ)}):=ψ(d*({circumflex over (θ)})−{tilde over (K)}({circumflex over (θ)}){tilde over (x)})

• • [0146] where ψ(s):=min{max{0, s},1} is a saturation function.

[0147] Then, we define the hybrid system ₀ with state z₀:=(q, τ)∈χ₀:={0,1}×[0,1], inputs x∈χ_P and

[0148] $\hat{\theta }\in {ℝ}_{>0}^2,$
and dynamics

[0149] ${\dot{z}}_0=F_0\left(z_0\right)\left(x,z_0,\hat{\theta }\right)\in C_0$ $z_0^{+}=G_0\left(z_0\right)\left(x,z_0,\hat{\theta }\right)\in D_0$ ${\kappa }_0\left(x,z_0,\hat{\theta }\right):=q$

• • [0150] where

[0151] $F_0\left(z_0\right):=\left[\begin{array}{c}0\\ 1/\epsilon \end{array}\right]$ $G_0\left(z_0\right):=\{\begin{array}{cc}\left[\begin{array}{c}0\\ \tau \end{array}\right]& \mathrm{if}\left(x,z_0,\theta \right)=D_0^0\\ \left[\begin{array}{c}1\\ 0\end{array}\right]& \mathrm{if}\left(x,z_0,\theta \right)=D_0^1\end{array}$ $\mathrm{with}{C}_0:={𝒳}_P\times {𝒳}_0\times {ℝ}_{>0}^2\mathrm{and}{D}_0:=D_0^0\bigcup D_0^1,\mathrm{where}$ $D_0^0:=\left\{\left(x,z_0,\hat{\theta }\right)\in {𝒳}_P\times {𝒳}_0\times {ℝ}_{>0}^2:\tau =d\left(x,\hat{\theta }\right),q=1\right\}$ $D_0^1:=\left\{\left(x,z_0,\hat{\theta }\right)\in {𝒳}_P\times {𝒳}_0\times {ℝ}_{>0}^2:\tau =1\right\}$

• • [0152] and ε>0 is a design parameter representing the PWM period.

[0153] For each solution (q, τ) to ₀, the state component r represents a timer that counts with a rate of 1/ε, and resets to zero each time τ=1. The state component q is a square wave representing the PWM signal that determines the converter switch state.

[0154] To ensure validity of the linearization in (17), and that the converter operates only in the continuous conduction mode under the local controller, we define the set χ_L:=∩Π. Then, since the matrix A_cl({circumflex over (θ)}) in (19) is Hurwitz for each {circumflex over (θ)}, there exists an open set ³_P⊂χ_L containing a neighborhood of _P that is forward invariant for [19]. The set _P is the basin of attraction for [19 14, Definition 7.3].

[0155] Appendix F establishes that every maximal solution to the closed-loop local controller is complete. Appendix G establishes convergence.

[0156] To implement the uniting control framework, the supervisor logic outlined in Example 3 is applied to the interconnection of the boost converter plant _P using the global and local control algorithms ₁ and ₀, respectively. Recall that z₀ is the state of ₀, z₁ is the state of ₁, and the output κ of the selected controller is mapped to the input q of _P. Then, we define the hybrid system _U with state z_U=(x, z₀, z₁, p)∈χ_U:=χ_P×χ₀×χ₁×{0,1} input

[0157] $\hat{\theta }\in {ℝ}_{>0}^2,$
and dynamics

[0158] $\begin{array}{cc}{\dot{z}}_U\in F_U\left(z_U,\hat{\theta }\right)\left(z_U,\hat{\theta }\right)\in C_U& \left(23\right)\end{array}$ $z_U^{+}=G_U\left(z_U\right)\left(z_U,\hat{\theta }\right)\in D_U.$

[0159] The logic variable p∈{0,1} is set to 0 when the global controller is selected and to 1 when the local controller is selected. The flow map F_U is equal to (F_P, F₀, 0,0) when p=0 and to (F_P, 0, F₁, 0) when p=1. It is written concisely as

[0160] $F_U\left(z_U,\hat{\theta }\right):=\left[\begin{array}{c}{F}_P\left(x,{\kappa }_p\left(x,z_p,\hat{\theta }\right)\right)\\ \left(1-p\right)F_0\left(z_0\right)\\ p{F}_1\left(z_1\right)\\ 0\end{array}\right]$

[0161] The flow set C_U is

[0162] $C_U:=\left\{\left(z_U,\hat{\theta }\right)\in {𝒳}_U\times R_{>0}^2:\left(x,{\kappa }_p\left(x,z_p,\hat{\theta }\right)\right)\in C_P\left(x,z_0,\hat{\theta }\right)\in C_0,\left(x,z_1,\hat{\theta }\right)\in C_1,\left(x,p\right)\in \left(\stackrel{\_}{ℳ}\times \left\{0\right\}\right)\bigcup \left(\stackrel{\_}{{ℝ}^n\backslash 𝒩}\times \left\{1\right\}\right)\right\}.$

• • [0163] where the sets and are to be designed.

[0164] The jump map G_U permits jumps by G₀ when p=0 and by G₁ when p=1, and toggles the value of p based on the converter state x in relation to the sets and . This is expressed as

[0165] $G_U\left(z_U\right):=\{\begin{array}{cc}{G}^0\left(z_U\right)& \left(z_U,\hat{\theta }\right)\in D^0\setminus D^2\\ G^1\left(z_U\right)& \left(z_U,\hat{\theta }\right)\in D^1\setminus D^2\\ G^2\left(z_U\right)& \left(z_U,\hat{\theta }\right)\in D^2\setminus \left(D^0\bigcup D^1\right)\\ \left\{G^0\left(z_U\right),G^2\left(z_U\right)\right\}& \left(z_U,\hat{\theta }\right)\in D^0\bigcap D^2\\ \left\{G^1\left(z_U\right),G^2\left(z_U\right)\right\}& \left(z_U,\hat{\theta }\right)\in D^1\bigcap D^2\end{array}$

• • [0166] where
    G⁰(z_U):=(x,G₀(z₀),z₁,p)
    G¹(z_U):=(x,z₀,G₁(z₁),p)
    G²(z_U):=(x,z₀,z₁,1−p)
  • [0167] and the jump set is D_U:=D⁰ ⊂D¹ ⊂D² with

[0168] $\begin{array}{c}{D}^0:=\left\{\left(z_U,\hat{\theta }\right)\in {𝒳}_U\times {ℝ}_{>0}^2:\left(x,z_0,\hat{\theta }\right)\in D_0,p=0\right\}\\ D^1:=\left\{\left(z_U,\hat{\theta }\right)\in {𝒳}_U\times {ℝ}_{>0}^2:\left(x,z_1,\hat{\theta }\right)\in D_1,p=1\right\}\\ D^2:=\left\{\left(z_U,\hat{\theta }\right)\in {𝒳}_U\times {ℝ}_{>0}^2:\left(x,p\right)\in \left(\stackrel{\_}{{ℝ}^n\backslash M}\times \left\{0\right\}\right)\bigcup \left(𝒩\times \left\{1\right\}\right)\right\}\end{array}$

[0169] The jump maps associated with the sets D⁰∩D² and D¹∩D² are necessary to satisfy outer semicontinuity of G in Definition 2.1.

[0170] Finally, we interconnect the uniting control algorithm _U in (23) and the estimation algorithm _E in (8) to obtain a hybrid system, denoted by , with state ξ:=(z_U,z_E)∈χ:=χ_U×χ_E and dynamics

[0171] $\begin{array}{cc}\dot{\xi }\in F\left(\xi \right)\xi \in C& \left(24\right)\end{array}$ ${\xi }^{+}=G\left(\xi \right)\xi \in D$

[0172] With

[0173] $F\left(\xi \right):=\left(F_U\left(z_U,\hat{\theta }\right),F_E\left(x,{\kappa }_p,z_E\right)\right)$ $C:=\left\{\xi \in \chi :\left(z_U,\hat{\theta }\right)\in C_U,\left(x,{\kappa }_p,z_E\right)\in C_E\right\}$ $G\left(\xi \right):=\{\begin{array}{cc}{G}_U\left(z_U\right),z_E)& \mathrm{if}\left(z_U,\hat{\theta }\right)\in D_U,\left(x,{\kappa }_p,z_E\right)\notin D_E\\ \left(z_U,G_E\left(z_E\right)\right)& \mathrm{if}\left(z_U,\hat{\theta }\right)\notin D_U,\left(x,{\kappa }_p,z_E\right)\in D_E\\ G_U\left(z_U\right),G_E\left(z_E\right))& \mathrm{if}\left(z_U,\hat{\theta }\right)\in D_U,\left(x,{\kappa }_p,z_E\right)\in D_E\end{array}$ $D:=\left\{\xi \in \chi :\left(z_U,\hat{\theta }\right)\in D_U\mathrm{or}\left(x,{\kappa }_p,z_E\right)\in D_E\right\}$

• • [0174] where we have omitted the arguments of κ_P for readability. Next we will design the uniting control sets and .

[0175] Any sets and that satisfy Assumption 3.1 in the appendix are acceptable for the uniting control framework in (23). We provide one example of how these sets can be designed for the boost converter. We define the closed set as

[0176] $𝒩:=x^{*}\left(\theta \right)+r_{𝒩}𝔹\mathrm{where}$ $r_{𝒩}\in {ℝ}_{>0}\mathrm{is}\mathrm{chosen}\mathrm{such}\mathrm{that}𝒩\subset {ℬ}_{{𝒜}_P}.$

• • [0177] Then, we choose ρ in (12) such that each maximal solution to the closed-loop global controller converges to . Note that since is centered on the set-point x*, it depends on the parameter estimate {circumflex over (θ)}. The choice of a ball for is arbitrary and was done for simplicity.

[0178] The reachable set from may be computed, for example, via Poisson analysis as in [20]. However, since this technique is computationally intensive for real-time implementation, we approximate M using the linearized model (19). A rigorous analysis of this approximation is beyond the scope of the paper. Using the Lyapunov function

[0179] $V\left(x\right):=x^{\top }Px,\mathrm{where}P=P^{\top }>0\mathrm{solves}$ $A_{\mathrm{cl}}^{\top }\left(\theta \right)P+{\mathrm{PA}}_{\mathrm{cl}}\left(\theta \right)=-Q\mathrm{and}Q=Q^{\top }>0,$
we choose a parameter r₀∈_>0 such that (r₀)⊃. Then, solutions to 19) from remain inside :=L_{{tilde over (V)}}(r₀)

[0180] To bound the trajectories of the closed-loop local controller, points on the boundary of are parameterized in a grid such that the variation in the vector field F_P between adjacent points is small. We compute the finite-time reachable set from each point on the boundary of by integrating F_P for ε seconds for each q∈{0,1}. Then, is defined as
:=int(L_{{tilde over (V)}}(r))  (26)

• • [0181] where r∈_>0 is chosen such that bounds the set reachable in ε seconds from for each switch state q∈{0,1}, as shown in FIG. 3A. Finally, we choose the matrix function {tilde over (K)} in 18) and the parameter ε>0 in (21) so that

[0182] $ℳ\subset {𝒳}_{ℒ}$

[0183] The following describes simulation results the hybrid system . Simulations are performed with c=0.1 F,L=0.2H,

[0184] $P=\left[\begin{array}{cc}c/2& 0\\ 0& L/2\end{array}\right],$
ε=0.0001, β=0.001, μ=0.001. The set in (25) is defined with r=0.05v_c*, and a grid of 10 points is used to compute in (26) from . Simulations are performed using the Hybrid Equations Toolbox [21].
(i) Simulating the Closed-Loop System

[0185] Simulation results for with {circumflex over (R)}(0,0)=R=3 and Ê(0,0)=E=5 are shown in FIGS. 3B and 3C (Code at https://github.com/HybridSystemsLab/UnitingBoost, which is incorporated by reference herein) for two separate initial conditions: x(0,0)=(5,0) and x(0,0)=(3,6). In both cases, the plant state converges to a neighborhood of _P=(7,3.27) in accordance with Theorem 5.3.

(ii) Simulation Results with Parameter Uncertainty

[0186] is simulated with uncertainty in the parameters R and E. Initial conditions are x(0,0)=(3,6), {circumflex over (R)}(0,0)=3, and Ê(0,0)=5. The parameters R and E are initially equal to 3.6 and 6, respectively. Then, E changes to 4 at t=3 seconds, and R changes to 2.4 at t=5 seconds. The parameter estimate converges to the true value at 0.5 seconds, as shown in FIG. 3D (Code at https://github.com/HybridSystemsLab/UnitingBoost, which is incorporated by reference herein) The estimate converges again after E changes at t=3, and after R changes at t=5. The plant state converges to a neighborhood of _P following each convergence of the parameter estimate to the true value, in accordance with Theorem 5.3 in the appendix

[0187] FIG. 4 is a flowchart illustrating a method of making a boost converter and/or control system for a boost converter.

[0188] Block 400 represents obtaining a boost converter circuit comprising a switch, a capacitor, an inductor, and a diode.

[0189] Block 402 represents connecting a hybrid adaptive controller circuit controlling a state (open or closed) of the switch so as to control raising of an input voltage E, inputted to the boost converter circuit, to an output voltage applied to an output load across the capacitor. The hybrid adaptive controller modulates a current flow through the inductor so as to control charging of the capacitor via the diode. The hybrid adaptive controller raises the input voltage to a desired output voltage by controlling the state of the switch

[0190] FIG. 2 illustrates the hybrid adaptive controller 200 comprises a parameter estimator 202 estimating the input voltage and the output load; and a converter estimator 204 estimating an output voltage across the capacitor using the input voltage and the output load, and the pulse width modulation as inputs. The controller further comprises a global controller 210 outputting a first signal that controls a state of the switch to converge the output voltage to the desired output voltage; and a local controller 208 outputting a second signal that controls the state of the switch to maintain the output voltage at the desired output voltage. The controller further comprises a supervisor 206 using the estimate of the output voltage as feedback to select:

• • [0191] the global controller outputting a signal (first switch state command) to control a state of the switch until the output voltage and/or inductor current have converged to within predetermined ranges.
  • [0192] the local controller to output a signal (second switch state command) controlling a state of the switch so long as the desired output voltage and/or inductor current are maintained within predetermined ranges.

[0193] Block 404 represents the end result, a boost converter system with hybrid adaptive control. Examples include, but are not limited to, the following (referring also to FIGS. 1-7).

[0194] 1. A boost converter 100, comprising:

• • [0195] a boost converter circuit 102 comprising a switch S (e.g., transistor) comprising a switch state (open or closed), a capacitor c, an inductor L, and a diode d;
  • [0196] a hybrid adaptive controller 104, 200, 602 connected to the switch for applying a hybrid algorithm controlling the switch state when the hybrid algorithm is executed by the hybrid adaptive controller; wherein the hybrid adaptive controller further comprises:
  • [0197] a global controller 210:
  • [0198] determining a first switch state command (open or close) as a function of estimates of the input voltage and a load resistance received from a parameter estimator, a measurement of an output voltage across the capacitor, and/or a measurement of the inductor current passing through the inductor, and
  • [0199] outputting the first switch state command (e.g., fixed state or changing with time) to the switch to set the switch state causing the output voltage to converge to a desired output voltage, thereby raising an input voltage E, inputted to the boost converter circuit, to the desired output voltage across the capacitor when a load comprising the load resistance is connected across the capacitor;
  • [0200] a local controller 208:
  • [0201] determining a second switch state command as a function of the estimates of the input voltage and the load resistance received from the parameter estimator, and the measurement of the output voltage and/or the measurement of the inductor current, and
  • [0202] outputting the second switch state command (e.g., pulse width modulation, fixed state or changing with time according to wave/signal) to the switch to set the switch state maintaining the output voltage at or within a set range of the desired output voltage and/or maintaining the inductor current at or within a set range of a desired current, when the load is connected; and
  • [0203] a supervisor 206 determining whether to execute the global controller outputting the first switch state command, or the local controller outputting the second switch state command, depending on the measurement of the output voltage and/or the measurement of the inductor current.

[0204] 2. The boost converter of example 1, wherein the hybrid adaptive controller further comprises the parameter estimator 202.

[0205] 3. The boost converter of example 1 or 2, wherein the supervisor uses the output voltage vs and/or the inductor current iL as feedback to select:

• • [0206] the global controller 210 outputting the switch state command until the output voltage has converged to within a first voltage range of the desired output voltage and/or the inductor current has converged to within a first current range of a desired current, or
  • [0207] the local controller 208 to output the second switch state command so long as the desired output voltage is maintained within a second voltage range and/or the inductor current is maintained within a second current range.

[0208] 4. The boost converter of any of the examples 1-3, wherein:

• • [0209] the global controller induces asymptotic stability of the output voltage at the desired output voltage and/or the inductor current at the desired current, and
  • [0210] the local controller maintains, in a the output voltage at the desired output voltage and/or the inductor current at the desired current.

[0211] 5. The boost converter of any of the examples 1-4, wherein the hybrid adaptive controller comprises one or more processors and one or more applications or programs executed by the one or more processors execute the hybrid algorithm comprising:

• • [0212] a parameter estimating algorithm, expressed using an estimator hybrid system, outputting the estimated load resistance and estimated input voltage in response to a parameter input comprising an initial input voltage;
  • [0213] a global control algorithm, expressed as a global controller hybrid system; outputting the first switch state command as a function of the output voltage and/or the inductor current,
  • [0214] a local control algorithm, expressed as a local controller hybrid system, outputting the second switch state command as a function of the output voltage and/or the inductor current; and
  • [0215] a supervisor algorithm, expressed as a supervisor hybrid system, outputting:
  • [0216] a first state instructing the hybrid adaptive controller to output the first switch state command to the switch until the output voltage is within a first voltage range and/or the inductor current is within a second current range, at which point outputting a second state, or
  • [0217] a second state instructing the hybrid controller to output the second switch state command so long as the output voltage is within a second voltage range and/or the inductor current is within a second current range.

[0218] 6. The boost converter of example 5, wherein the first state and the second state comprise logic states 0 or 1.

[0219] 7. The boost converter of example 5, wherein:

• • [0220] the parameter estimating algorithm is initialized with at least one of the current or the voltage set to zero, so that the first set of values have converged to the desired output voltage and/or desired current, for any value and polarity of the desired output voltage or the desired current.

[0221] 8. The boost converter of example 5, wherein the one or more applications or programs obtain or create a representation of the hybrid systems and determine the switch state commands by processing, in the hybrid systems, state variables representing the switch state commands as a function of the output voltage and/or desired current.

[0222] 9. The boost converter of example 8, wherein the processing of the state variables comprises modifying the state variables via discrete jumps and continuous evolution in the hybrid systems.

[0223] 10. The boost converter of example 8 or 9, wherein the hybrid systems determine the switch state commands from the estimates of the load resistance and the input voltage, and the measurements of the output voltage and the current, using ohm's law relationships between the inductor current, the output voltage, the input voltage, and the load resistance.

[0224] 11. The boost converter of any of the examples 1-10, wherein:

• • [0225] the first voltage range and the first current range each comprise a first set of values to which maximal solutions of the output voltage and the inductor current, respectively, converge in finite time in response to the first switch state command, and
  • [0226] the second voltage range and the second current range each comprise a second set of values M to which the output voltage and the inductor current, respectively, remain in steady state in response to the second switch state command.

[0227] 12. The boost converter of example 11, wherein the first set of values N comprises positive values of the output voltage across the capacitor and/or positive values of a current passing through the inductor.

[0228] 13. The boost converter of any of the examples 1-12, wherein the local controller hybrid system implements a timer.

[0229] 14. The boost converter of any of the examples 1-13, wherein at least one of the global controller, local controller, or the supervisor comprise one or more circuits comprising one or more application specific integrated circuits or one or more field programmable gate arrays.

[0230] 15. The boost converter of any of the examples, wherein the parameter estimator is expressed using an estimator hybrid system.

[0231] 16. The boost converter of any of the examples 1-15, wherein the controller comprises one or more processors; one or more memories; and one or more applications or programs stored in the one or more memories, wherein the one or more applications or programs executed by the one or more processors execute one or more algorithms, comprising:

• • [0232] a parameter estimating algorithm, expressed using an estimator hybrid system in the parameter estimator, outputting estimated load resistance and estimated input voltage in response to a parameter input comprising an initial input voltage;
  • [0233] a voltage estimating algorithm, expressed as a converter hybrid system in the convertor estimator 204, outputting the estimate of the output voltage, using the estimated load resistance and the estimated input voltage, in response to a converter input comprising the pulse modulation signal (e.g., whether the switch is on or off),
  • [0234] a global control algorithm, expressed as a global controller hybrid system in the global controller 210; outputting a first commanded switch state as a function of the output voltage,
  • [0235] a local control algorithm, expressed as a local controller hybrid system in the local controller 208, outputting the second commanded switch state as a function of the output voltage; and
  • [0236] a supervisor algorithm, expressed as a supervisor hybrid system in the supervisor 206, outputting:
  • [0237] a first state instructing the controller to output the first modulation signal to the switch until the estimate is within the first range, at which point outputting the second state, or
  • [0238] a second state instructing the controller to output the second modulation signal so long as the estimate is within the second range.

[0239] 17. A boost converter, comprising:

• • [0240] a boost converter circuit comprising a switch, a capacitor, an inductor, and a diode;
  • [0241] an adaptive controller applying a switch state command to the switch so as to control:
  • [0242] raising of an input voltage E, inputted to the boost converter circuit, to a desired output voltage applied to an output load across the capacitor, and/or
  • [0243] flow of the current through the inductor, wherein the switch state modulates a flow of the current through the inductor so as to control charging of the capacitor via the diode; and
  • [0244] the hybrid adaptive controller comprises:
  • [0245] a parameter estimator estimating the input voltage and the output load;
  • [0246] a converter estimator estimating at least one of an output voltage across the capacitor, or the current through the inductor, using the input voltage, the output load, and the pulse width modulation as inputs;
  • [0247] a global controller outputting a first switch state command to converge at least one of the output voltage to the desired output voltage, or the current to a desired current;
  • [0248] a local controller outputting a second switch state command to maintain the output voltage at the desired output voltage, or the current at a desired current; and
  • [0249] a supervisor using the estimate of the output voltage or the current as feedback to select:
  • [0250] the global controller outputting the first switch state command until the output voltage or the current has converged to within a first voltage range of the desired output voltage, and/or to within a first current range of the desired current, respectively, or
  • [0251] the local controller to output the second switch state command so long as the desired output voltage or the current is maintained within a second voltage range and/or second current range, respectively.

[0252] 18. The boost converter of any of the examples, wherein the controller comprises one or more processors and one or more applications or programs executed by the one or more processors execute one or more algorithms, comprising:

• • [0253] a parameter estimating algorithm, expressed using an estimator hybrid system, outputting estimated load resistance and estimated input voltage in response to a parameter input comprising an initial input voltage;
  • [0254] an estimating algorithm, expressed as a converter hybrid system, outputting the estimate of the output voltage and/or the current, using the estimated load resistance and the estimated input voltage, in response to a converter input comprising the pulse modulation signal,
  • [0255] a global control algorithm, expressed as a global controller hybrid system; outputting the first modulation signal as a function of the estimate of the output voltage and/or the current,
  • [0256] a local control algorithm, expressed as a local controller hybrid system, outputting the second modulation signal as a function of the estimate of the output voltage and/or the current; and
  • [0257] a supervisor algorithm, expressed as a supervisor hybrid system, outputting:
  • [0258] a first state instructing the controller to output the first switch state command to the switch until the voltage estimate is within the first voltage range and/or the current estimate is within the second current range, at which point outputting the second state, or
  • [0259] a second state instructing the controller to output the second switch state command so long as the voltage estimate is within the second voltage range and/or the current estimate is within the second current range.

[0260] 19. A computer implemented method for controlling a boost converter, comprising:

• • [0261] obtaining a boost converter circuit comprising a switch comprising a switch state (open or closed), a capacitor, an inductor, and a diode;
  • [0262] applying a hybrid algorithm controlling the switch state, comprising:
  • [0263] using a global controller:
  • [0264] determining a first switch state command (open or close) as a function of estimates of the input voltage and a load resistance received from a parameter estimator, a measurement of an output voltage across the capacitor, and a measurement of the inductor current passing through the inductor, and
  • [0265] outputting the first switch state command to the switch to set the switch state causing the output voltage to converge to a desired output voltage, thereby raising an input voltage E, inputted to the boost converter circuit, to the desired output voltage across the capacitor when a load comprising the load resistance is connected across the capacitor;
  • [0266] using a local controller:
  • [0267] determining a second switch state command as a function of the estimates of the input voltage and the load resistance received from the parameter estimator, and the measurement of the output voltage and the measurement of the inductor current, and
  • [0268] outputting the second switch state command to the switch to set the switch state maintaining the output voltage at or within a set range of the desired output voltage and maintaining the inductor current at or within a set range of a desired current, when the load is connected; and
  • [0269] determining, in a supervisor, whether to execute the global controller outputting the first switch state command, or the local controller outputting the second switch state command, depending on the measurement of the output voltage and the measurement of the inductor current.

[0270] 20. A computer implemented system 600 for controlling a boost converter circuit, comprising:

• • [0271] one or more processors 604A-B, or a computer readable medium configured for:
  • [0272] receiving at least one of an desired output voltage or an inductor current for application to an output load using the boost converter circuit comprising a switch comprising a switch state (open or closed), a capacitor, an inductor, and a diode;
  • [0273] determining a first switch state command (open or closed) as a function of estimates of an input voltage and a load resistance received from a parameter estimator, a measurement of an output voltage across the capacitor, and a measurement of the inductor current passing through the inductor, wherein the first switch state command sets the switch state causing the output voltage to converge to a desired output voltage, thereby raising an input voltage E, inputted to the boost converter circuit, to the desired output voltage across the capacitor when a load comprising the load resistance is connected across the capacitor; and
  • [0274] determining a second switch state command as a function of the estimates of the input voltage and the load resistance received from the parameter estimator, and the measurement of the output voltage and the measurement of the inductor current, wherein the second switch state command sets the switch state maintaining the output voltage at or within a set range of the desired output voltage and/or maintains the inductor current at or within a set range of a desired current, when the load is connected; and
  • [0275] determining whether to output the first switch state command or the second switch state command to control the switch, depending on the measurement of the output voltage and the measurement of the inductor current.

[0276] 21. The system of example 20 comprising the one or more processors and one or more applications or programs executed by the one or more processors for executing a hybrid algorithm performing the determining steps and outputting the switch state commands.

[0277] 22. The system of example 20 or 21, wherein at least one of the global controller, local controller, or the supervisor comprise one or more circuits comprising one or more application specific integrated circuits or one or more field programmable gate arrays.

[0278] 23. The system or method of any of the examples 19-22 implemented using the circuit of any of the examples 1-18.

[0279] 24. A control system for a boost converter circuit comprising a model that captures the transient behavior and every possible state of the converter system; a modular “uniting” control framework (e.g., as described in example 3) that switches between global and local control schemes, implemented using a hybrid system tool (e.g., as presented in example 2); and an estimator (e.g., as in example 3) permitting finite-time estimation of the converter input voltage and load resistance. Example 5 shows that the closed-loop uniting framework and finite-time estimator induce global practical asymptotic stability of a desired voltage value. The hybrid systems tools developed in [14,15] form the enabling techniques to achieve these results.

[0280] 25. A control system comprising an estimator that permits finite-time estimation of the converter input voltage and load resistance; a modular uniting framework/module that combines global and local control algorithms, and global and local control module implementing a global and local control scheme that induces the desired stability properties, wherein the closed-loop adaptive control scheme induces global practical asymptotic stability of the set-point while offering industry-standard PWM control performance near the set point even under parameter uncertainty. This is a substantial/significant improvement over PWM control alone.

[0281] 26. A control system comprising a parameter estimator estimating input voltage and load resistance; a global controller outputting a first switch state command (fixed state or changing state with time) to converge the output voltage to desired output voltage and determining switch state using estimated input voltage, estimated load resistance from parameter estimator, and measured output voltage and measured inductor current, and a hybrid system analysis; and local controller outputting second switch state command (fixed state or varying switch state with time) to maintain the output voltage at the desired output voltage and determining switch state using estimated input voltage, estimated load resistance from parameter estimator, and measured output voltage and measured inductor current, and a hybrid system analysis.

[0282] 27. The controller of any of the examples, wherein the processing performed by each controller (global controller, local controller, supervisor) is an algorithm, a method/methodology, or a function or one or more circuits. The global controller and the local controller, estimator, and supervisor may each comprise one or more components, modules, systems, units, etc. The combined set of elements within the controller and each controller may processes input using one or more algorithms or circuits to generate/produce an output.

[0283] 28. The boost converter or system of any of the examples wherein the controller 104, 200 (e.g., comprising global controller, local controller, supervisor) comprises one or more processors; one or more memories; and one or more applications or programs stored in the one or more memories, wherein the one or more applications or programs executed by the one or more processors execute the one or more algorithms described herein.

[0284] 29. The boost converter can be a converter circuit that converts direct current (DC) to DC.

[0285] In one or more examples, the controller 104 can also be used to control the output voltage by lowering or raising or otherwise changing the input voltage.

[0286] FIG. 5 illustrates a method of controlling a boost converter circuit.

[0287] Block 500 represents obtaining a desired output voltage for a boost converter circuit comprising a switch, a capacitor, an inductor, and a diode.

[0288] Block 502 represents determining and applying, using the output voltage as feedback, a hybrid algorithm for controlling raising of an input voltage E, inputted to the boost converter circuit, to an output voltage comprising the desired output voltage applied to an output load across the capacitor. The hybrid algorithm modulates a current flow through the inductor so as to control charging of the capacitor via the diode.

[0289] As illustrated in FIG. 2, the controlling comprises:

• • [0290] estimating, using a parameter estimator hybrid system, an estimated load resistance and estimated input voltage in response to an input comprising an initial input voltage,
  • [0291] controlling the output voltage and/or current, using a converter hybrid system receiving the estimated load resistance, the estimated input voltage and the switch state commands as inputs.
  • [0292] determining, using a global controller hybrid system, a first commanded switch state (for controlling a switch state of the switch) as a function of the output voltage and the current,
  • [0293] determining, using a local controller hybrid system, a second commanded switch state (for controlling a switch state of the switch, e.g. using pulse width modulation) as a function of the output voltage and current,
  • [0294] selecting, using a supervisor hybrid system:
  • [0295] the global controller hybrid system to output the first commanded switch state until the output voltage has converged to within a first range of the desired output voltage and/or current, or
  • [0296] the local controller hybrid system to output the second commanded switch state so long as the desired output voltage and/or current is maintained within a second range.

[0297] Block 504 represents obtaining the desired output voltage/inductor current.

[0298] FIG. 6 is an exemplary hardware and software environment 600 (referred to as a computer-implemented system and/or computer-implemented method) used to implement one or more embodiments of the invention. The hardware and software environment includes a computer 602 and may include peripherals. Computer 602 may be a user/client computer, server computer, or may be a database computer. The computer 602 comprises a hardware processor 604A and/or a special purpose hardware processor 604B (hereinafter alternatively collectively referred to as processor 604) and a memory 606, such as random access memory (RAM). The computer 602 may be coupled to, and/or integrated with, other devices, including input/output (I/O) devices such as a keyboard 614, a cursor control device 616 (e.g., a mouse, a pointing device, pen and tablet, touch screen, multi-touch device, etc.) and a printer 628. In one or more embodiments, computer 602 may be coupled to, or may comprise, a portable or media viewing/listening device 632 (e.g., an MP3 player, IPOD, NOOK, portable digital video player, cellular device, personal digital assistant, etc.). In yet another embodiment, the computer 602 may comprise a multi-touch device, mobile phone, gaming system, internet enabled television, television set top box, or other internet enabled device executing on various platforms and operating systems.

[0299] In one embodiment, the computer 602 operates by the hardware processor 604A performing instructions defined by the computer program 610 (e.g., hybrid adaptive control application) under control of an operating system 608. The computer program 610 and/or the operating system 608 may be stored in the memory 606 and may interface with the user and/or other devices to accept input and commands and, based on such input and commands and the instructions defined by the computer program 610 and operating system 608, to provide output and results.

[0300] Output/results may be presented on the display 622 or provided to another device for presentation or further processing or action. In one embodiment, the display 622 comprises a liquid crystal display (LCD) having a plurality of separately addressable liquid crystals. Alternatively, the display 622 may comprise a light emitting diode (LED) display having clusters of red, green and blue diodes driven together to form full-color pixels. Each liquid crystal or pixel of the display 622 changes to an opaque or translucent state to form a part of the image on the display in response to the data or information generated by the processor 604 from the application of the instructions of the computer program 610 and/or operating system 608 to the input and commands. The image may be provided through a graphical user interface (GUI) module 618. Although the GUI module 618 is depicted as a separate module, the instructions performing the GUI functions can be resident or distributed in the operating system 608, the computer program 610, or implemented with special purpose memory and processors.

[0301] In one or more embodiments, the display 622 is integrated with/into the computer 602 and comprises a multi-touch device having a touch sensing surface (e.g., track pod or touch screen) with the ability to recognize the presence of two or more points of contact with the surface. Examples of multi-touch devices include mobile devices (e.g., IPHONE, NEXUS S, DROID devices, etc.), tablet computers (e.g., IPAD, HP TOUCHPAD, SURFACE Devices, etc.), portable/handheld game/music/video player/console devices (e.g., IPOD TOUCH, MP3 players, NINTENDO SWITCH, PLAYSTATION PORTABLE, etc.), touch tables, and walls (e.g., where an image is projected through acrylic and/or glass, and the image is then backlit with LEDs).

[0302] Some or all of the operations performed by the computer 602 according to the computer program 610 instructions may be implemented in a special purpose processor 604B. In this embodiment, some or all of the computer program 610 instructions may be implemented via firmware instructions stored in a read only memory (ROM), a programmable read only memory (PROM) or flash memory within the special purpose processor 604B or in memory 606. The special purpose processor 604B may also be hardwired through circuit design to perform some or all of the operations to implement the present invention. Further, the special purpose processor 604B may be a hybrid processor, which includes dedicated circuitry for performing a subset of functions, and other circuits for performing more general functions such as responding to computer program 610 instructions. In one embodiment, the special purpose processor 604B is an application specific integrated circuit (ASIC).

[0303] The computer 602 may also implement a compiler 612 that allows an application or computer program 610 written in a programming language such as C, C++, Assembly, SQL, PYTHON, PROLOG, MATLAB, RUBY, RAILS, HASKELL, or other language to be translated into processor 604 readable code. Alternatively, the compiler 612 may be an interpreter that executes instructions/source code directly, translates source code into an intermediate representation that is executed, or that executes stored precompiled code. Such source code may be written in a variety of programming languages such as JAVA, JAVASCRIPT, PERL, BASIC, etc. After completion, the application or computer program 610 accesses and manipulates data accepted from I/O devices and stored in the memory 606 of the computer 602 using the relationships and logic that were generated using the compiler 612.

[0304] The computer 602 also optionally comprises an external communication device such as a modem, satellite link, Ethernet card, or other device for accepting input from, and providing output to, other computers 602.

[0305] In one embodiment, instructions implementing the operating system 608, the computer program 610, and the compiler 612 are tangibly embodied in a non-transitory computer-readable medium, e.g., data storage device 620, which could include one or more fixed or removable data storage devices, such as a zip drive, floppy disc drive 624, hard drive, CD-ROM drive, tape drive, etc. Further, the operating system 608 and the computer program 610 are comprised of computer program 610 instructions which, when accessed, read and executed by the computer 602, cause the computer 602 to perform the steps necessary to implement and/or use the present invention or to load the program of instructions into a memory 606, thus creating a special purpose data structure causing the computer 602 to operate as a specially programmed computer executing the method steps described herein. Computer program 610 and/or operating instructions may also be tangibly embodied in memory 606 and/or data communications devices 630, thereby making a computer program product or article of manufacture according to the invention. As such, the terms “article of manufacture,” “program storage device,” and “computer program product,” as used herein, are intended to encompass a computer program accessible from any computer readable device or media.

[0306] Of course, those skilled in the art will recognize that any combination of the above components, or any number of different components, peripherals, and other devices, may be used with the computer 602.

[0307] FIG. 7 schematically illustrates a typical distributed/cloud-based computer system 700 using a network 704 to connect client computers 702 to server computers 706. A typical combination of resources may include a network 704 comprising the Internet, LANs (local area networks), WANs (wide area networks), SNA (systems network architecture) networks, or the like, clients 702 that are personal computers or workstations (as set forth in FIG. 6), and servers 706 that are personal computers, workstations, minicomputers, or mainframes (as set forth in FIG. 6). However, it may be noted that different networks such as a cellular network (e.g., GSM [global system for mobile communications] or otherwise), a satellite based network, or any other type of network may be used to connect clients 702 and servers 706 in accordance with embodiments of the invention.

[0308] A network 704 such as the Internet connects clients 702 to server computers 706. Network 704 may utilize ethernet, coaxial cable, wireless communications, radio frequency (RF), etc. to connect and provide the communication between clients 702 and servers 706. Further, in a cloud-based computing system, resources (e.g., storage, processors, applications, memory, infrastructure, etc.) in clients 702 and server computers 706 may be shared by clients 702, server computers 706, and users across one or more networks. Resources may be shared by multiple users and can be dynamically reallocated per demand. In this regard, cloud computing may be referred to as a model for enabling access to a shared pool of configurable computing resources.

[0309] Clients 702 may execute a client application or web browser and communicate with server computers 706 executing web servers 710. Such a web browser is typically a program such as MICROSOFT INTERNET EXPLORER/EDGE, MOZILLA FIREFOX, OPERA, APPLE SAFARI, GOOGLE CHROME, etc. Further, the software executing on clients 702 may be downloaded from server computer 706 to client computers 702 and installed as a plug-in or ACTIVEX control of a web browser. Accordingly, clients 702 may utilize ACTIVEX components/component object model (COM) or distributed COM (DCOM) components to provide a user interface on a display of client 702. The web server 710 is typically a program such as MICROSOFT'S INTERNET INFORMATION SERVER.

[0310] Web server 710 may host an Active Server Page (ASP) or Internet Server Application Programming Interface (ISAPI) application 712, which may be executing scripts. The scripts invoke objects that execute business logic (referred to as business objects). The business objects then manipulate data in database 716 through a database management system (DBMS) 714. Alternatively, database 716 may be part of, or connected directly to, client 702 instead of communicating/obtaining the information from database 716 across network 704. When a developer encapsulates the business functionality into objects, the system may be referred to as a component object model (COM) system. Accordingly, the scripts executing on web server 710 (and/or application 712) invoke COM objects that implement the business logic. Further, server 706 may utilize MICROSOFT'S TRANSACTION SERVER (MTS) to access required data stored in database 716 via an interface such as ADO (Active Data Objects), OLE DB (Object Linking and Embedding DataBase), or ODBC (Open DataBase Connectivity).

[0311] Generally, these components 700-716 all comprise logic and/or data that is embodied in/or retrievable from device, medium, signal, or carrier, e.g., a data storage device, a data communications device, a remote computer or device coupled to the computer via a network or via another data communications device, etc. Moreover, this logic and/or data, when read, executed, and/or interpreted, results in the steps necessary to implement and/or use the present invention being performed.

[0312] Although the terms “user computer”, “client computer”, and/or “server computer” are referred to herein, it is understood that such computers 702 and 706 may be interchangeable and may further include thin client devices with limited or full processing capabilities, portable devices such as cell phones, notebook computers, pocket computers, multi-touch devices, and/or any other devices with suitable processing, communication, and input/output capability.

[0313] Of course, those skilled in the art will recognize that any combination of the above components, or any number of different components, peripherals, and other devices, may be used with computers 702 and 706. Embodiments of the invention are implemented as a software/hybrid adaptive control application on a client 702 or server computer 706. Further, as described above, the client 702 or server computer 706 may comprise a thin client device or a portable device that has a multi-touch-based display.

[0314] The present disclosure solves the problem of practically asymptotically stabilizing the DC-DC boost converter under parameter uncertainty. The industry-standard PWM control algorithm for the boost converter uses a linearized averaged model of the converter dynamics to determine the control input. As a result, the PWM control scheme guarantees stability only in the region of the state space where the linearization holds.

[0315] The present disclosure describes an estimation algorithm that identifies the input voltage and output load of the converter in finite time. Using these estimates, we designed a control algorithm that “unites” global and local control schemes. The global control scheme induces practical asymptotic stability of a desired output voltage and corresponding current, and the local control scheme maintains industry-standard PWM behavior during steady state. Stability properties for the resulting hybrid closed-loop system are established and operation is demonstrated through simulation, showing that by combining global and local control algorithms leveraging hybrid system tools, the hybrid adaptive controller induces practical asymptotic stability of a desired set point from a larger set of initial conditions than the PWM control algorithm.

[0316] Although the control architecture was demonstrated using hybrid system, alternative control schemes for the global or local control algorithms in the uniting control framework may be used. The control algorithm may also be implemented in hardware.

2.1. Notation

[0317] We denote the real, nonnegative, positive, and natural numbers as , _≥0, _>0, and , respectively. Given a set S, ∂S denotes its boundary and S its closure. The Euclidean norm is denoted |⋅|. The distance of a point x to a nonempty set S is defined by |x|_S=inf_y∈s|y−x|. Given a set-valued mapping M:^mn, the domain of M is the set dom M={x∈^m:M(x)≠∅}, and the range of M is the set rge M={y∈ⁿ:∃x∈^m s.t. y∈M(x)}. A function β:_≥0×_≥0→_≥0 is said to be of class if it is nondecreasing in its first argument, nonincreasing in its second argument, lim_r→0+β(r, s)=0 for each s∈_≥0, and lim_s→∞β(r, s)=0 for each r∈_≥0. The (Bouligand) tangent cone [14. Definition 5.12] to the set S⊂ⁿ at η∈ⁿ is denoted T_S(η). The μ-sublevel set of the function V: dom V→_≥0, which is the set of points {x∈dom V:V(x)≤μ}, is denoted L_V (μ). The closed unit ball centered at the origin of appropriate dimension (in the Euclidean norm) is denoted . The set contains positive semidefinite matrices with dimension n×n.

2.1 Hybrid System

[0318] An example hybrid system has data (C, F, D, G, κ) and is given by [14,15]

[0319] $\mathscr{H}=\{\begin{array}{ccc}\left(x,u\right)\in C& \dot{x}& \in F\left(x,u\right)\\ \left(x,u\right)\in D& x^{+}& \in G\left(x,u\right)\\ y& & =\kappa \left(x,u\right)\end{array}$

• • [0320] where x∈ⁿ is the state, u∈^m is the input, F:ⁿ×^{m n} is a setvalued map defining the flow map of the differential inclusion capturing the continuous dynamics, and C⊂ⁿ×^m defines the flow set on which flows are permitted. Similarly, G:ⁿ×^mn is a set-valued map defining the jump map of the difference inclusion modeling the discrete behavior, and D⊂ⁿ×^m is the jump set on which jumps are permitted. The vector y∈ⁿ defines the output of the hybrid system.

[0321] A solution x to is parameterized by (t,j)∈_≥0×, where t is the amount of ordinary time that has passed and j is the number of jumps that have occurred. The domain of x, denoted dom x⊂_≥0×, is a hybrid time domain, in the sense that for every (T,J)∈dom x, there exists a nondecreasing sequence

[0322] ${\left\{t_j\right\}}_{j=0}^{J+1}$
with t₀=0 such that

[0323] $\mathrm{dom}x\cap \left(\left[0,T\right]\times \left\{0,1,\dots ,J\right\}\right)=\bigcup _{j=0}^J\left(\left[t_j,t_{j+1}\right],\left\{j\right\}\right)$

[0324] A solution x to is called maximal if it cannot be extended. A solution is called complete if its domain is unbounded. The set of all maximal solutions to is denoted , and the set of all maximal solutions to with initial condition belonging to a set K is denoted (K). A set K is said to be forward invariant for if each solution x∈z,671 (K) is complete and satisfies rge x⊂K. We will use [14, Proposition 6.10] to prove the existence and completeness of solutions for the proposed algorithm. For self-contained-ness, we recall 14, Proposition 6.10] next.

[0325] Definition 2.1. Let =(C, F, D, G) satisfy the hybrid basic conditions, i.e., its data (C, F, D, G) is such that (A1) C and D are closed subsets of ⁿ; (A2) F:ⁿⁿ is outer semicontinuous and locally bounded relative to C, C⊂dom F, and F(x) is nonempty and convex for all x∈C;

• • [0326] (A3) G:ⁿⁿ is outer semicontinuous and locally bounded relative to D and D⊂dom G.

[0327] Take an arbitrary ζ∈C∪D. If ζE D, or (VC) there exists a neighborhood U of ζ such that for every x∈U∩C, F(x)∩T_C(x)≠∅ then there exists a nontrivial solution to with x(0,0)=ζ. If (VC) holds for every ζ∈C\D, then there exists a nontrivial solution to H from every point in C∪D, and every x∈ satisfies exactly one of the following conditions:

• • [0328] (a) x is complete;
  • [0329] (b) (finite escape time) dom x is bounded and the interval I^J, where J=sup_jdom x, has nonempty interior and tx(t,J) is a maximal solution to {dot over (x)}∈F(x), in fact lim_t>T|x(t,J)|=∞, where T=sup_tdom x;
  • [0330] (c) (solutions jumping outside C∪D)dom x(T,J)∉C∪D, where (T,J)=sup dom x.

[0331] Furthermore, if G(D)⊂C∪D, then (c) above does not occur.

2.2 Boost Converter Model

[0332] FIG. 1 illustrates a boost converter comprising is a class of switched-mode power supply that utilizes a switch S, inductor L, diode d, and capacitor c to raise the voltage at the output load R compared to the input voltage E. The state of the switch S (open or closed) represents the control input to the boost converter plant. When the switch is closed, current flows through the inductor and generates a magnetic field. When the switch is opened, the inductor magnetic field decays to maintain the current towards the load, causing a polarity reversal within the inductor. The primary voltage source in series with the inductor then charges the capacitor through the diode to a higher voltage than is attainable using the voltage source alone. If the switch is cycled fast enough, the inductor does not fully discharge between cycles and the load voltage remains higher than that of the source.

[0333] The boost converter dynamics may be expressed as a (continuous time) plant, _P, with discrete-valued input denoting the position of the switch S. We model it as in (1) but with no jumps. That is, _P with state

[0334] $x:=\left(v_c,i_L\right),𝓍\in {𝒳}_P:=\underset{\_}{{\tilde{M}}_0^{⎵}}\bigcup {\tilde{M}}_1$
which are defined below, input q∈{0, 1},
and output x. Following 12], its dynamics reduce to the differential inclusion with
constraints

[0335] $x\in F_P\left(x,q\right)\left(x,q\right)\in C_P\mathrm{where}$ $C_P:=\left({\tilde{M}}_0\times \left\{0\right\}\right)\bigcup \left({\tilde{M}}_1\times \left\{1\right\}\right)$ ${\tilde{M}}_0:=\stackrel{\_}{M_1}\bigcup \stackrel{\_}{M_3}=\left\{x\in {ℝ}^2:i_L\ge 0\right\},$ ${\tilde{M}}_1:=M_2\left\{x\in {ℝ}^2:v_c\ge 0\right\},\mathrm{with}$ $M_1=\left\{x\in {ℝ}^2:i_L>0\right\}\bigcup \left\{x\in {ℝ}^2:v_c\le E,i_L=0\right\}$ $M_2=\left\{x\in {ℝ}^2:v_c\ge 0\right\}$ $M_3=\left\{x\in {ℝ}^2:v_c>E,i_L=0\right\}$

• • [0336] and F_P is the Krasovskii regularization of the vector fields and the corresponding constraint sets associated with each mode of circuit operation given by, for each x∈χ_P, (see [12 for details)

[0337] $F_P\left(x,0\right):=\{\begin{array}{cc}\left[\begin{array}{c}-\frac{1}{\mathrm{Rc}}{v}_c+\frac{1}{c}{i}_L\\ -\frac{1}{L}{v}_c+\frac{E}{L}\end{array}\right]& \mathrm{if}x\in \stackrel{\_}{M_1}\setminus \stackrel{\_}{M_3}\\ \left\{-\frac{1}{\mathrm{Rc}}{v}_c\right\}\times \left[-\frac{1}{L}{v}_c+\frac{E}{L},0\right]& \mathrm{if}x\in \stackrel{\_}{M_3}\end{array}$ $F_P\left(x,1\right):=\begin{array}{cc}\left[\begin{array}{c}-\frac{1}{\mathrm{Rc}}{v}_c\\ \frac{E}{L}\end{array}\right]& \mathrm{if}x\in \widetilde{M_1}\end{array}.$

[0338] Since the supply voltage E and load resistance R for the converter may vary during operation, it is desirable to estimate these values so the chosen control algorithm can adapt accordingly. Furthermore, since the stability properties of boost converter PWM control algorithms typically only hold locally near the set-point, we desire to solve the following problems:

• • [0339] Problem 1: Design an algorithm that permits on-line estimation of the boost converter input voltage E and load resistance R.
  • [0340] Problem 2: Design an adaptive control law whose closed-loop system induces practical asymptotic stability of a desired output voltage.

[0341] Due to our desire to maintain PWM operation in steady-state, the boost converter is an ideal candidate for the control framework known as “uniting control.” This framework utilizes a divide and conquer approach to control design by combining two feedback controllers and a logic-based algorithm that selects which controller to apply. Uniting control strategies permit combining a global controller that renders a set-point stable but may not have good performance near the set point, and a local controller that induces satisfactory performance, but only locally [13].

[0342] The uniting control logic is implemented as follows. Given a plant _P as in (2) interconnected with two separate control algorithms, ₀, referred to as “global,” and ₁, referred to as “local,” the choice of control algorithm is governed by a supervisory controller that selects between ₁ and ₀ based on the plant state in relation to a closed set and an open set ⊃as follows: —Apply the global controller ₁ until the solution to the plant enters . When any such point is reached, switch to the local controller ₀.

• • [0343] Apply the local controller ₀ as long as the solution to the plant remains inside . If the state of the plant leaves , switch to the global controller ₁.

[0344] Given ₁ and ₀, the uniting control sets and will be designed to satisfy the following assumption.

[0345] Assumption 3.1. Given a hybrid system _P as in (2),

• • [0346] each maximal solution x to _P with input q generated by ₁ converges to in finite (hybrid) time;
  • [0347] each solution x to _P from and input q generated by ₁ remains in for all (hybrid) time.

[0348] Hence, for the boost converter, the global controller drives the converter state into , from where solutions under the local PWM controller remain inside for all future time.

[0349] Next, the parameter estimation algorithm and the global and local control algorithms are designed and then combined using the uniting control framework.

[0350] Next we address Problem 1 from Section 3. Suppose that the unknown converter parameters R and E take values from R∈[R_min, R_max] and E∈[E_min, E_max] where R_min, R_max, E_min, E_max>0 are known. Then, for the purpose of estimating R and E, we establish the following lemma that allows us to express the dynamics of maximal solutions to _P, in a convenient form.

[0351] Lemma 4.1. Each maximal solution tx(t) to _P in (2) with input tq(t) satisfie {
{dot over (x)}(t)=f₁(x(t),q(t))+f₂(x(t),q(t))ϑ)  (4)

[0352] Since _P in (2) is a continuous-time system, its solutions are parameterized using only t. for all t∈dom (x, q), where ϑ=(ϑ₁,ϑ₂):=(R⁻¹, E) and

[0353] Proof. This proof is in Appendix A.

[0354] Estimating the parameters R and E is equivalent to estimating the parameter vector ϑ in (4). For this purpose, the finite-time parameter estimator in [16,17] is extended to classes of hybrid systems whose solutions satisfy 4). The algorithm is expressed as a hybrid system, denoted _E, and operates as follows. Let z_E=({circumflex over (x)}, {circumflex over (θ)}, ω, Q, η, Γ) be a solution to _E—hence, defined on a hybrid time domain—where {circumflex over (x)} is the estimate of x, {circumflex over (ϑ)} is the estimate of ϑ, and ω, Q, η, Γ are auxiliary variables. Consider the initial interval of flow I⁰:={t:(t, 0)∈dom z_E} in dom z_E with constant ϑ and initial conditions ω(0,0)=0,Q(0,0)=0,η(0,0)=0,Γ(0,0)=0, and {circumflex over (ϑ)}(0,0)∈² arbitrary. Omitting the (t,j) of solutions, for the sake of making an argument, suppose that over this interval of flow, Q and Γ satisfy
{dot over (Q)}=ω^τω,{dot over (Γ)}=ω^τωϑ
Then, if there exists a positive time t₁∈I⁰ such that Q(t₁, 0) is invertible, resetting {circumflex over (ϑ)} to the value of Q⁻¹Γ leads to

[0355] $\begin{array}{cc}\hat{\vartheta }\left(t_1,1\right)=Q^{-1}\left(t_1,0\right)\Gamma \left(t_1,0\right)& \left(5\right)\end{array}$ $\begin{array}{cc}={\left({\int }_0^{t_1}{\omega \left(t,0\right)}^{\top }\omega \left(t,0\right)\mathrm{dt}\right)}^{-1}\left({\int }_0^{t_1}{\omega \left(t,0\right)}^{\top }\omega \left(t,0\right)\mathrm{dt}\vartheta \right)=\vartheta & \left(6\right)\end{array}$

[0356] However, since ϑ is unknown prior to hybrid time (t₁, 1), a trajectory for Γ satisfying (5) cannot be generated. Due to this, we rewrite the dynamics of Γ as

[0357] $\begin{array}{c}\dot{\Gamma }={\omega }^{\top }\left(\mathrm{\omega \vartheta }+\omega \left(\hat{\vartheta }-\hat{\vartheta }\right)+\left(x-x\right)+\left(\hat{x}-\hat{x}\right)\right)\\ ={\omega }^{\top }\left(\omega \hat{\vartheta }+x-\hat{x}-\eta \right)\end{array}$

• • [0358] where η=x−{circumflex over (x)}−ω(ϑ−{circumflex over (ϑ)}). Note that since ω(0,0)=0, the initial condition η(0,0)=0 implies {circumflex over (x)}(0,0)=x(0,0). Differentiating η yields
    {dot over (η)}={dot over (x)}−{circumflex over ({dot over (x)})}−{dot over (ω)}(ϑ−{circumflex over (ϑ)})+ω{circumflex over ({dot over (ϑ)})}

[0359] Next we define a matrix function (x, q)K(x, q)=K^τ(x, q)>0 that is to be designed. The arguments of K are omitted below for simplicity. Let {circumflex over (x)}, ω, and {circumflex over (ϑ)} satisfy
{circumflex over ({dot over (x)})}=f₁(x,q)+f₂(x,q){circumflex over (ϑ)}+K(x−{circumflex over (x)})+ω{circumflex over ({dot over (ϑ)})}
{dot over (ω)}=f₂(x,q)−Kω.

[0360] Plugging the expressions above into (7) yields

[0361] $\begin{array}{cc}\dot{\eta }=f_2\left(x,q\right)\left(\vartheta -\hat{\vartheta }\right)-K\left(x-\hat{x}\right)-\omega \stackrel{.}{\stackrel{ˆ}{\vartheta }}-\left(f_2\left(x,q\right)-K\omega \right)\left(\vartheta -\hat{\vartheta }\right)+\omega \stackrel{.}{\stackrel{ˆ}{\vartheta }}=-K\left(x-\hat{x}\right)+K\omega \left(\vartheta -\hat{\vartheta }\right)=-K\eta & \left(7\right)\end{array}$

[0362] Hence, ω, Q, η, and Γ are now expressed in terms of known quantities and we can compute {circumflex over (ϑ)} as in (6).

[0363] Following [17], the estimation scheme outlined above can be implemented as a hybrid algorithm whose jump map imposes the initial conditions specified above (5) and computes {circumflex over (ϑ)} as in (6). The hybrid system, denoted by _E=(C_E, F_E, D_E, G_E, {circumflex over (θ)}), has state

[0364] $z_E:=\left(\hat{x},\hat{\vartheta },\omega ,Q,\eta ,\Gamma \right)\in {𝒳}_E:={ℝ}^2\times {ℝ}^2\times {ℝ}^{2\times 2}\times {\mathrm{𝒮𝒫}}^{2\times 2}\times {ℝ}^2\times {ℝ}^2,$ $\mathrm{inputs}\left(x,q\right)\in C_P,\mathrm{output}\hat{\theta }\in {ℝ}_{>0}^2,$
and dynamics

[0365] $\begin{array}{cc}{\stackrel{.}{z}}_E=F_E\left(x,q,z_E\right)\left(x,q,z_E\right)\in C_E& \left(8\right)\end{array}$ $z_E^{+}=G_E\left(x,z_E\right)\left(x,q,z_E\right)\in D_E$ $\hat{\theta }=h_E\left(\hat{\vartheta }\right)\mathrm{where}{G}_E\left(x,z_E\right):=\left(x,Q^{-1}\Gamma ,0,0,0,0\right),$ $F_E\left(x,q,z_E\right):=\left[\begin{array}{c}{f}_1\left(x,q\right)+f_2\left(x,q\right)\hat{\vartheta }+K\left(x-\hat{x}\right)+\omega h\left(x,q,z_E\right)\\ h\left(x,q,z_E\right)\\ f_2\left(x,q\right)-K\omega \\ {\omega }^{\top }\omega \\ -K\eta \\ {\omega }^{\top }\left(\omega \hat{\vartheta }+x-\hat{x}-\eta \right)\end{array}\right]\mathrm{with}$ $h\left(x,q,z_E\right)=\Omega \left({\omega }^{\top }+{f_2\left(x,q\right)}^{\top }\right)\left(x-\hat{x}\right),\mathrm{and}$ $C_E:=\left\{\left(x,q,z_E\right)\in C_P\times {𝒳}_E:\det \left(Q\right)\le \mu \right\}$ $D_E:=\left\{\left(x,q,z_E\right)\in C_P\times {𝒳}_E:\det \left(Q\right)\ge \mu \right\}.$