Intelligent mechatronic control suspension system based on soft computing

[0001] 1. Field of the Invention

[0002] The disclosed invention is relates generally to control systems, and more particularly to electronically controlled suspension systems.

[0003] 2. Description of the Related Art

[0004] Feedback control systems are widely used to maintain the output of a dynamic system at a desired value in spite of external disturbances that would displace it from the desired value. For example, a household space-heating furnace, controlled by a thermostat, is an example of a feedback control system. The thermostat continuously measures the air temperature inside the house, and when the temperature falls below a desired minimum temperature the thermostat turns the furnace on. When the interior temperature reaches the desired minimum temperature, the thermostat turns the furnace off. The thermostat-furnace system maintains the household temperature at a substantially constant value in spite of external disturbances such as a drop in the outside temperature. Similar types of feedback controls are used in many applications.

[0005] A central component in a feedback control system is a controlled object, a machine, or a process that can be defined as a plant, having an output variable or performance characteristic to be controlled. In the above example, the plant is the house, the output variable is the interior air temperature in the house and the disturbance is the flow of heat (dispersion) through the walls of the house. The plant is controlled by a control system. In the above example, the control system is the thermostat in combination with the furnace. The thermostat-furnace system uses simple on-off feedback control system to maintain the temperature of the house. In many control environments, such as motor shaft position or motor speed control systems, simple on-off feedback control is insufficient. More advanced control systems rely on combinations of proportional feedback control, integral feedback control, and derivative feedback control. A feedback control based on a sum of proportional feedback, plus integral feedback, plus derivative feedback, is often referred as PID control.

[0006] A PID control system is a linear control system that is based on a dynamic model of the plant. In classical control systems, a linear dynamic model is obtained in the form of dynamic equations, usually ordinary differential equations. The plant is assumed to be relatively linear, time invariant, and stable. However, many real-world plants are time-varying, highly non-linear, and unstable. For example, the dynamic model may contain parameters (e.g., masses, inductance, aerodynamics coefficients, etc.), which are either only approximately known or depend on a changing environment. If the parameter variation is small and the dynamic model is stable, then the PID controller may be satisfactory. However, if the parameter variation is large or if the dynamic model is unstable, then it is common to add adaptive or intelligent (AI) control functions to the PID control system.

[0007] AI control systems use an optimizer, typically a non-linear optimizer, to program the operation of the PID controller and thereby improve the overall operation of the control system.

[0008] Classical advanced control theory is based on the assumption that near of equilibrium points all controlled plants can be approximated as linear systems. Unfortunately, this assumption is rarely true in the real world. Most plants are highly nonlinear, and often do not have simple control algorithms. In order to meet these needs for a nonlinear control, systems have been developed that use soft computing concepts such as genetic algorithms, fuzzy neural networks, fuzzy controllers and the like. By these techniques, the control system evolves (changes) over time to adapt itself to changes that may occur in the controlled plant and/or in the operating environment.

[0009] Currently, self-organizing control systems based on fuzzy controllers suffer from two drawbacks. First, when a genetic analyzer is used to develop a teaching signal for a fuzzy neural network, the teaching signal typically contains unnecessary stochastic noise, making it difficult to later develop an approximation to the teaching signal. Second, the fitness functions used for the genetic analyzers in self-organizing suspension control systems typically optimize the control system according to some desired control paradigm without reference to human factors such as rider comfort.

[0010] The present invention solves these and other problems by providing a control system for optimizing a shock absorber system having a non-linear kinetic characteristic. The control system uses a fitness (performance) function that is based on the physical laws of minimum entropy and biologically inspired constraints relating to rider comfort, driveability, etc. In one embodiment, a genetic analyzer is used in an off-line mode to develop a teaching signal. An information filter is used to filter the teaching signal to produce a compressed teaching signal. The compressed teaching signal can be approximated online by a fuzzy controller that operates using knowledge from a knowledge base. The control system can be used to control complex plants described by nonlinear, unstable, dissipative models. The control system is configured to use smart simulation techniques for controlling the shock absorber (plant).

[0011] In one embodiment, the control system comprises a learning system, such as a neural network that is trained by a genetic analyzer. The genetic analyzer uses a fitness function that maximizes sensor information while minimizing entropy production based on biologically-inspired constraints.

[0012] In one embodiment, a suspension control system uses a difference between the time differential (derivative) of entropy (called the production entropy rate) from the learning control unit and the time differential of the entropy inside the controlled process (or a model of the controlled process) as a measure of control performance. In one embodiment, the entropy calculation is based on a thermodynamic model of an equation of motion for a controlled process plant that is treated as an open dynamic system.

[0013] The control system is trained by a genetic analyzer that generates a teaching signal. The optimized control system provides an optimum control signal based on data obtained from one or more sensors. For example, in a suspension system, a plurality of angle and position sensors can be used. In an off-line learning mode (e.g., in the laboratory, factory, service center, etc.), fuzzy rules are evolved using a kinetic model (or simulation) of the vehicle and its suspension system. Data from the kinetic model is provided to an entropy calculator that calculates input and output entropy production of the model. The input and output entropy productions are provided to a fitness function calculator that calculates a fitness function as a difference in entropy production rates for the genetic analyzer constrained by one or more constraints obtained from rider preferences. The genetic analyzer uses the fitness function to develop a training signal for the off-line control system. The training signal is filtered to produce a compressed training signal. Control parameters from the off-line control system are then provided to an online control system in the vehicle that, using information from a knowledge base, develops an approximation to the compressed training signal.

[0014] In one embodiment, the invention includes a method for controlling a nonlinear object (a plant) by obtaining an entropy production difference between a time differentiation (dS_u/dt) of the entropy of the plant and a time differentiation (dS_c/dt) of the entropy provided to the plant from a controller. A genetic algorithm that uses the entropy production difference as a fitness (performance) function evolves a control rule in an off-line controller. The nonlinear stability characteristics of the plant are evaluated using a Lyapunov function. The genetic analyzer minimizes entropy and maximizes sensor information content. Filtered control rules from the off-line controller are provided to an online controller to control suspension system. In one embodiment, the online controller controls the damping factor of one or more shock absorbers (dampers) in the vehicle suspension system.

[0015] In some embodiments, the control method also includes evolving a control rule relative to a variable of the controller by means of a genetic algorithm. The genetic algorithm uses a fitness function based on a difference between a time differentiation of the entropy of the plant (dS_u/dt) and a time differentiation (dS_c/dt) of the entropy provided to the plant. The variable can be corrected by using the evolved control rule.

[0016] In one embodiment, the invention comprises a self-organizing control system adapted to control a nonlinear plant. The AI control system includes a simulator configured to use a thermodynamic model of a nonlinear equation of motion for the plant. The thermodynamic model is based on a Lyapunov function (V), and the simulator uses the function V to analyze control for a state stability of the plant. The control system calculates an entropy production difference between a time differentiation of the entropy of said plant (dS_u/dt) and a time differentiation (dS_c/dt) of the entropy provided to the plant by a low-level controller that controls the plant. The entropy production difference is used by a genetic algorithm to obtain an adaptation function wherein the entropy production difference is minimized in a constrained fashion. The genetic algorithm provides a teaching signal. The teaching signal is filtered to remove stochastic noise to produce a filtered teaching signal. The filtered teaching signal is provided to a fuzzy logic classifier that determines one or more fuzzy rules by using a learning process. The fuzzy logic controller is also configured to form one or more control rules that set a control variable of the controller in the vehicle.

[0017] In yet another embodiment, the invention comprises a new physical measure of control quality based on minimum production entropy and using this measure for a fitness function of genetic algorithm in optimal control system design. This method provides a local entropy feedback loop in the control system. The entropy feedback loop provides for optimal control structure design by relating stability of the plant (using a Lyapunov function) and controllability of the plant (based on production entropy of the control system). The control system is applicable to a wide variety of control systems, including, for example, control systems for mechanical systems, bio-mechanical systems, robotics, electro-mechanical systems, etc.

[0018] The above and other aspects, features, and advantages of the present invention will be more apparent from the following description thereof presented in connection with the following drawings.

[0019] FIG. 1 illustrates a general structure of a self-organizing intelligent control system based on soft computing.

[0020] FIG. 2 illustrates the structure of a self-organizing intelligent suspension control system with physical and biological measures of control quality based on soft computing

[0021] FIG. 3 illustrates the process of constructing the Knowledge Base (KB) for the Fuzzy Controller (FC).

[0022] FIG. 4 shows twelve typical road profiles.

[0023] FIG. 5 shows a normalized auto-correlation function for different velocities of motion along the road number 9 from FIG. 4.

[0024] FIG. 6A is a plot showing results of stochastic simulations based on a one-dimensional Gaussian probability density function.

[0025] FIG. 6B is a plot showing results of stochastic simulations based on a one-dimensional uniform probability density function.

[0026] FIG. 6C is a plot showing results of stochastic simulations based on a one-dimensional Rayleigh probability density function.

[0027] FIG. 6D is a plot showing results of stochastic simulations based on a two-dimensional Gaussian probability density function.

[0028] FIG. 6E is a plot showing results of stochastic simulations based on a two-dimensional uniform probability density function.

[0029] FIG. 6F is a plot showing results of stochastic simulations based on a two-dimensional hyperbolic probability density function.

[0030] FIG. 7 illustrates a full car model.

[0031] FIG. 8A shows a control damper layout for a suspension-controlled vehicle having adjustable dampers.

[0032] FIG. 8B shows an adjustable damper for the suspension-controlled vehicle.

[0033] FIG. 8C shows fluid flow for soft and hard damping in the adjustable damper from FIG. 8B.

[0034] FIG. 9 shows damper force characteristics for the adjustable dampers illustrated in FIG. 8.

[0035] FIG. 10 shows the structure of an SSCQ for use in connection with a simulation model of the full car and suspension system.

[0036] FIG. 11 is a flowchart showing operation of the SSCQ.

[0037] FIG. 12 shows time intervals associated with the operating mode of the SSCQ.

[0038] FIG. 13 is a flowchart showing operation of the SSCQ in connection with the GA.

[0039] FIG. 14 shows the genetic analyzer process and the operations of reproduction, crossover, and mutation.

[0040] FIG. 15 shows results of variables for the fuzzy neural network.

[0041] FIG. 16A shows control of a four-wheeled vehicle using two controllers.

[0042] FIG. 16B shows control of a four-wheeled vehicle using a single controller to control all four wheels.

[0043] FIG. 17 shows an information filter application results for compression of a normal distribution stochastic signal, including: (a) the source signal; (b) the row-wise information distribution before compression; (c) the signal after compression; and (d) the information distribution after compression.

[0044] FIG. 18 shows the teaching signal K^c provided by the information filter 241 shown in FIG. 2.

[0045] FIG. 19 shows the output of the simulation system of control quality of intelligent control suspension system.

[0046] FIG. 20 shows the simulation results of fuzzy control of a suspension system

[0047] FIG. 21 shows the square of heave jerk for controlled and uncontrolled systems.

[0048] FIG. 22 shows the square of pitch jerk for controlled and uncontrolled systems.

[0049] FIG. 23 shows the square of roll jerk for controlled and uncontrolled systems.

[0050] In the drawings, the first digit of any three-digit element reference number generally indicates the number of the figure in which the referenced element first appears. The first two digits of any four-digit element reference number generally indicate the figure in which the referenced element first appears.

[0051] FIG. 1 is a block diagram of a control system 100 for controlling a plant based on soft computing. In the controller 100, a reference signal y is provided to a first input of an adder 105. An output of the adder 105 is an error signal , which is provided to an input of a Fuzzy Controller (FC) 143 and to an input of a Proportional-Integral-Differential (PID) controller 150. An output of the PID controller 150 is a control signal u*, which is provided to a control input of a plant 120 and to a first input of an entropy-calculation module 132. A disturbance m(t) 110 is also provided to an input of the plant 120. An output of the plant 120 is a response x, which is provided to a second input the entropy-calculation module 132 and to a second input of the adder 105. The second input of the adder 105 is negated such that the output of the adder 105 (the error signal ) is the value of the first input minus the value of the second input.

[0052] An output of the entropy-calculation module 132 is provided as a fitness function to a Genetic Analyzer (GA) 131. An output solution from the GA 131 is provided to an input of a FNN 142. An output of the FNN 142 is provided as a knowledge base to the FC 143. An output of the FC 143 is provided as a gain schedule to the PID controller 150.

[0053] The GA 131 and the entropy calculation module 132 are part of a Simulation System of Control Quality (SSCQ) 130. The FNN 142 and the FC 143 are part of a Fuzzy Logic Classifier System (FLCS) 140.

[0054] Using a set of inputs, and the fitness function 132, the genetic algorithm 131 works in a manner similar to a biological evolutionary process to arrive at a solution which is, hopefully, optimal. The genetic algorithm 131 generates sets of chromosomes (that is, possible solutions) and then sorts the chromosomes by evaluating each solution using the fitness function 132. The fitness function 132 determines where each solution ranks on a fitness scale. Chromosomes (solutions) which are more fit are those chromosomes which correspond to solutions that rate high on the fitness scale. Chromosomes which are less fit are those chromosomes which correspond to solutions that rate low on the fitness scale.

[0055] Chromosomes that are more fit are kept (survive) and chromosomes that are less fit are discarded (die). New chromosomes are created to replace the discarded chromosomes. The new chromosomes are created by crossing pieces of existing chromosomes and by introducing mutations.

[0056] The PID controller 150 has a linear transfer function and thus is based upon a linearized equation of motion for the controlled plant 120. Prior art genetic algorithms used to program PID controllers typically use simple fitness and thus do not solve the problem of poor controllability typically seen in linearization models. As is the case with most optimizers, the success or failure of the optimization often ultimately depends on the selection of the performance (fitness) function.

[0057] Evaluating the motion characteristics of a nonlinear plant is often difficult, in part due to the lack of a general analysis method. Conventionally, when controlling a plant with nonlinear motion characteristics, it is common to find certain equilibrium points of the plant and the motion characteristics of the plant are linearized in a vicinity near an equilibrium point. Control is then based on evaluating the pseudo (linearized) motion characteristics near the equilibrium point. This technique is scarcely, if at all, effective for plants described by models that are unstable or dissipative.

[0058] Computation of optimal control based on soft computing includes the GA 131 as the first step of global search for an optimal solution on a fixed space of positive solutions. The GA searches for a set of control weights for the plant. Firstly the weight vector Kk₁, . . . , k_n is used by a conventional proportional-integral-differential (PID) controller 150 in the generation of a signal u*(K) which is applied to the plant. The entropy S((K)) associated to the behavior of the plant 120 on this signal is used as a fitness function by the GA 131 to produce a solution that gives minimum entropy production. The GA 131 is repeated several times at regular time intervals in order to produce a set of weight vectors K. The vectors K generated by the GA 131 are then provided to the FNN 142 and the output of the FNN 142 to the fuzzy controller 143. The output of the fuzzy controller 143 is a collection of gain schedules for the PID controller 150 that controls the plant. For the soft computing system 100 based on a genetic analyzer, there is very often no real control law in the classical control sense, but rather, control is based on a physical control law such as minimum entropy production.

[0059] In order to realize an intelligent mechatronic suspension control system, the structure depicted on FIG. 1 is modified, as shown on FIG. 2 to produce a system 200 for controlling a plant, such as suspension system. The system 200 is similar to the system 100 with the addition of an information filter 241 and biologically-inspired constraints 233 in the fitness function 132. The information filter 241 is placed between the GA 131 and the FNN 142 such that a solution vector output K_i from the GA 131 is provided to an input of the information filter 241. An output of the information filter 241 is a filtered solution vector K_c that is provided to the input of the FNN 142. In FIG. 2, the disturbance 110 is a road signal m(t). (e.g., measured data or data generated via stochastic simulation). In FIG. 2, the plant 120 is a suspension system and car body. The fitness function 132, in addition to entropy production rate, includes biologically-inspired constraints based on mechanical and/or human factors. In one embodiment, the filter 241 includes an information compressor that reduces unnecessary noise in the input signal of the FNN 142. In FIG. 2, the PID controller 150 is shown as a proportional damping force controller.

[0060] As shown in FIG. 3, realization of the structure depicted in FIG. 2 is divided into four development stages. The development stages include a teaching signal acquisition stage 301, a teaching signal compression stage 302, a teaching signal approximation stage 303, and a knowledge base verification stage 304.

[0061] The teaching signal acquisition stage 301 includes the acquisition of a robust teaching signal without the loss of information. In one embodiment, the stage 301 is realized using stochastic simulation of a full car with a Simulation System of Control Quality (SSCQ) under stochastic excitation of a road signal. The stage 301 is based on models of the road, of the car body, and of models of the suspension system. Since the desired suspension system control typically aims for the comfort of a human, it is also useful to develop a representation of human needs, and transfer these representations into the fitness function 132 as constraints 233.

[0062] The output of the stage 301 is a robust teaching signal K_i, which contains information regarding the car behavior and corresponding behavior of the control system.

[0063] Behavior of the control system is obtained from the output of the GA 131, and behavior of the car is a response of the model for this control signal. Since the teaching signal K_i is generated by a genetic algorithm, the teaching signal K_i typically has some unnecessary stochastic noise in it. The stochastic noise can make it difficult to realize (or develop a good approximation for) the teaching signal K_i. Accordingly, in a second stage 302, the information filter 241 is applied to the teaching signal K_i to generate a compressed teaching signal K_c. The information filter 241 is based on a theorem of Shannon's information theory (the theorem of compression of data). The information filter 241 reduces the content of the teaching signal by removing that portion of the teaching signal K_i that corresponds to unnecessary information. The output of the second stage 302 is a compressed teaching signal K_c.

[0064] The third stage 303 includes approximation of the compressed teaching signal K_c by building a fuzzy inference system using a fuzzy logic classifier (FLC) based on a Fuzzy Neural Network (FNN). Information of car behavior can be used for training an input part of the FNN, and corresponding information of controller behavior can be used for output-part training of the FNN.

[0065] The output of the third stage 303 is a knowledge base (KB) for the FC 143 obtained in such a way that it has the knowledge of car behavior and knowledge of the corresponding controller behavior with the control quality introduced as a fitness function in the first stage 301 of development. The KB is a data file containing control laws of the parameters of the fuzzy controller, such as type of membership functions, number of inputs, outputs, rule base, etc.

[0066] In the fourth stage 304, the KB can be verified in simulations and in experiments with a real car, and it is possible to check its performance by measuring parameters that have been optimized.

[0067] To summarize, the development of the KB for an intelligent control suspension system includes:

[0068] I. Obtaining a stochastic model of the road or roads.

[0069] II. Obtaining a realistic model of a car and its suspension system.

[0070] III. Development of a Simulation System of Control Quality with the car model for genetic algorithm fitness function calculation, and introduction of human needs in the fitness function.

[0071] IV. Development of the information compressor (information filter).

[0072] V. Approximation of the teaching signal with a fuzzy logic classifier system (FLCS) and obtaining the KB for the FC

[0073] VI. Verification of the KB in experiment and/or in simulations of the full car model with fuzzy control

[0074] I. Obtaining Stochastic Models of the Roads

[0075] It is convenient to consider different types of roads as stochastic processes with different auto-correlation functions and probability density functions. FIG. 4 shows twelve typical road profiles. Each profile shows distance along the road (on the x-axis), and altitude of the road (on the y-axis) with respect to a reference altitude. FIG. 5 shows a normalized auto-correlation function for different velocities of motion along the road number 9 (from FIG. 4). In FIG. 5, a curve 501 and a curve 502 show the normalized auto-correlation function for a velocity 1 meter/sec, a curve 503 shows the normalized auto-correlation function for 5 meter/sec, and a curve 504 shows the normalized auto-correlation function for 10 meter/sec.

[0076] The results of statistical analysis of actual roads, as shown in FIG. 4, show that it is useful to consider the road signals as stochastic processes using the following three typical auto-correlation functions.

R()B(0)exp₁cos ₁;(1.2) $\begin{array}{cc}R\left(\right)=B\left(0\right)\exp \left\{-{}_1\left|\right|\right\}\left[\cos {}_1+\frac{{}_1}{{}_1}\sin \left({}_1\left|\right|\right)\right],& \left(1.3\right)\end{array}$

[0077] where ₁ and ₁ are the values of coefficients for single velocity of motion. The ranges of values of these coefficients are obtained from experimental data as:

[0078] ₁0.014 to 0.111; ₁0.025 to 0.140.

[0079] For convenience, the roads are divided into three classes:

[0080] A. {square root}{square root over (B(0))}10 smsmall obstacles;

[0081] B. {square root}{square root over (B(0))}10 sm to 20 smmedium obstacles;

[0082] C. {square root}{square root over (B(0))}>20 smlarge obstacles.

[0083] The presented auto-correlation functions and its parameters are used for stochastic simulations of different types of roads using forming filters. The methodology of forming filter structure can be described according to the first type of auto-correlation functions (1.1) with different probability density functions.

[0084] Consider a stationary stochastic process X(t) defined on the interval x_l,x_r, which can be either bounded or unbounded. Without loss of generality, assume that X(t) has a zero mean. Then x_l<0 and x_r>0. With the knowledge of the probability density p(x) and the spectral density _XX() of X(t), one can establish a procedure to model the process X(t).

[0085] Let the spectral density be of the following low-pass type: $\begin{array}{cc}{}_{\mathrm{XX}}\left(\right)=\frac{{}^2}{\left({}^2+{}^2\right)},>0,& \left(2.1\right)\end{array}$

[0086] where ² is the mean-square value of X(t). If X(t) is also a diffusive Markov process, then it is governed by the following stochastic differential equation in the Ito sense (see Appendix I):

[0087] where is the same parameter in (2.1), B(t) is a unit Wiener process, and the coefficients X and D(X) are known as drift and the diffusion coefficients, respectively. To demonstrate that this is the case, multiply (2.2) by X(t) and take the ensemble average to yield $\begin{array}{cc}\frac{R\left(\right)}{}=-R\left(\right),& \left(2.3\right)\end{array}$

[0088] where R() is the correlation function of X(t), namely, R()EX(t)X(t). Equation (2.3) has a solution

[0089] in which A is arbitrary. By choosing A², equations (2.1) and (2.4) become a Fourier transform pair. Thus equation (2.2) generates a process X(t) with a spectral density (2.1). Note that the diffusion coefficient D(X) has no influence on the spectral density.

[0090] Now it is useful to determine D(X) so that X(t) possesses a given stationary probability density p(x). The Fokker-Planck equation, governing the probability density p(x) of X(t) in the stationary state, is obtained from equation (2.2) as follows: $\begin{array}{cc}\frac{}{x}G=-\frac{}{x}\left\{\mathrm{xp}\left(x\right)+\frac{1}{2}\frac{}{x}\left[D^2\left(x\right)p\left(x\right)\right]\right\}=0,& \left(2.5\right)\end{array}$

[0091] where G is known as the probability flow. Since X(t) is defined on x_l,x_r, G must vanish at the two boundaries xx_l and xx_r. In the present one-dimensional case, G must vanish everywhere; consequently, equation (2.5) reduces to $\begin{array}{cc}\mathrm{xp}\left(x\right)+\frac{1}{2}\frac{}{x}\left[D^2\left(x\right)p\left(x\right)\right]=0.& \left(2.6\right)\end{array}$

[0092] Integration of equation (2.6) results in $\begin{array}{cc}{D}^2\left(x\right)p\left(x\right)=-2{}_{x_l}^{x_r}\mathrm{up}\left(u\right)u+C,& \left(2.7\right)\end{array}$

[0093] where C is an integration constant. To determine the integration constant C, two cases are considered. For the first case, if x_l, or x_r, or both, then p(x)must vanish at the infinite boundary; thus C0 from equation (2.7). For the second case, if both x_l and x_r are finite, then the drift coefficient x_l at the left boundary is positive, and the drift coefficient x_r at the right boundary is negative, indicating that the average probability flows at the two boundaries are directed inward. However, the existence of a stationary probability density implies that all sample functions must remain within x_l,x_r, which requires additionally that the drift coefficient vanish at the two boundaries, namely, D²(x_l)D²(x_r)0. This is satisfied only if C0. In either case, $\begin{array}{cc}{D}^2\left(x\right)=-\frac{2}{p\left(x\right)}{}_{x_l}^{x_r}\mathrm{up}\left(u\right)u.& \left(2.8\right)\end{array}$

[0094] Function D²(x), computed from equation (2.8), is non-negative, as it should be, since p(x)0 and the mean value of X(t) is zero. Thus the stochastic process X(t) generated from (2.2) with D(x) given by (2.8) possesses a given stationary probability density p(x) and the spectral density (2.1).

[0095] The Ito type stochastic differential equation (2.2) may be converted to that of the Stratonovich type (see Appendix I) as follows: $\begin{array}{cc}\stackrel{.}{X}=-X-\frac{1}{4}\frac{D^2\left(X\right)}{X}+\frac{D\left(X\right)}{\sqrt{2}}\left(t\right),& \left(2.9\right)\end{array}$

[0096] where (t) is a Gaussian white noise with a unit spectral density. Equation (2.9) is better suited for simulating sample functions. Some illustrative examples are given below.

[0097] Example 1: Assume that X(t) is uniformly distributed, namely $\begin{array}{cc}p\left(x\right)=\frac{1}{2},-x.& \left(2.10\right)\end{array}$

[0098] Substituting (2.10) into (2.8)

[0099] In this case, the desired Ito equation is given by

[0100] It is of interest to note that a family of stochastic processes can be obtained from the following generalized version of (2.12):

[0101] Their appearances are strikingly diverse, yet they share the same spectral density (2.1).

[0102] Example 2: Let X(t) be governed by a Rayleigh distribution

[0103] Its centralized version Y(t)X(t)2/ has a probability density

[0104] From equation (2.8), $\begin{array}{cc}{D}^2\left(y\right)=\frac{2}{}\left(y+\frac{2}{}\right).& \left(2.16\right)\end{array}$

[0105] The Ito equation for Y(t) is $\begin{array}{cc}\mathrm{dY}=-\mathrm{Ydt}+{\left[\frac{2}{}\left(Y+\frac{2}{}\right)\right]}^{1/2}dB\left(t\right)& \left(2.17\right)\end{array}$

[0106] and the correspondence equation for X(t) in the Stratonovich form is $\begin{array}{cc}\stackrel{.}{X}=-X+\frac{3}{2}+{\left(\frac{}{}X\right)}^{1/2}\left(t\right).& \left(2.18\right)\end{array}$

[0107] Note that the spectral density of X(t) contains a delta function (4/²)() due to the nonzero mean 2/.

[0108] Example 3: Consider a family of probability densities, which obeys an equation of the form $\begin{array}{cc}\frac{}{x}p\left(x\right)=J\left(x\right)p\left(x\right).& \left(2.19\right)\end{array}$

[0109] Equation (2.19) can be integrated to yield

[0110] where C₁ is a normalization constant. In this case

[0111] Several special cases may be noted. Let

[0112] where can be arbitrary if >0. Substitution of equation (2.22) into equation (2.8) leads to $\begin{array}{cc}{D}^2\left(x\right)=\frac{}{2}\sqrt{/}\exp \left[{\left(x^2+\frac{}{2}\right)}^2\right]\mathrm{erfc}\left[\sqrt{}\left(x^2+\frac{}{2}\right)\right],& \left(2.23\right)\end{array}$

[0113] where erfc(y) is the complementary error function defined as $\begin{array}{cc}\mathrm{erfc}\left(y\right)=\frac{2}{\sqrt{}}{}_y^{}{e}^{-t^2}\mathrm{dt}.& \left(2.24\right)\end{array}$

[0114] The case of <0 and >0 corresponds to a bimodal distribution, and the case of >0 and 0 corresponds to a Gaussian distribution.

[0115] The Pearson family of probability distributions corresponds to $\begin{array}{cc}J\left(x\right)=\frac{a_{1}x+a_0}{b_2{x}^2+b_{1}x+b_0}& \left(2.25\right)\end{array}$

[0116] In the special case of a₀b₁0, $\begin{array}{cc}{D}^2\left(x\right)=-\frac{2}{a_1+b_2}\left(b_2{x}^2+b_{1}x+b_0\right).& \left(2.26\right)\end{array}$

[0117] From the results of statistical analysis of forming filters with auto-correlation function (1.1) one can describe typical structure of forming filters as in Table 2.1:

TABLE 2.1
The Structures of Forming Filters for Typical Probability Density Functions p(x)
Auto-correlation	Probability
function	density function	Forming filter structure
R y ( ) = 2 -	Gaussian	y . + y = 2 ( t )
R y ( ) = 2 -	Uniform	y . + 2 y = 2 2 ( 2 - y 2 ) ( t )
R y ( ) = 2 -	Rayleigh	y . + y 2 = 2 2 2 ( y + 2 ) ( t )
R y ( ) = 2 -	Pearson	y . + y + a 1 + 2 b 2 ( b 2 x + b 1 ) = 2 2 2 a 1 + 2 b 2 ( b 2 y 2 + b 1 y + b 0 )

[0118] The structure of a forming filter with an auto-correlation function given by equations (1.2) and (1.3) is derived as follows. A two-dimensional (2D) system is used to generate a narrow-band stochastic process with the spectrum peak located at a nonzero frequency. The following pair of Ito equations describes a large class of 2D systems:

[0119] where B_i, i1,2 are two independent unit Wiener processes.

[0120] For a system to be stable and to possess a stationary probability density, is required that a₁₁<0, a₂₂<0 and a₁₁a₂₂a₁₂a₂₁>0. Multiplying (3.1) by x₁(t)and taking the ensemble average, gives $\begin{array}{cc}\frac{}{}{R}_{11}\left(\right)=a_{11}{R}_{11}\left(\right)+a_{12}{R}_{12}\left(\right)\frac{}{}{R}_{12}\left(\right)=a_{21}{R}_{11}\left(\right)+a_{22}{R}_{12}\left(\right)& \left(3.2\right)\end{array}$

[0121] where R₁₁()Mx₁(t)x₁(t), R₁₂()Mx₁(t)x₂(t) with initial conditions R₁₁(0)m₁₁Mx₁², R₁₂(0)m₁₂Mx₁x₂.

[0122] Differential equations (3.2) in the time domain can be transformed (using the Fourier transform) into algebraic equations in the frequency domain as follows $\begin{array}{cc}i{\stackrel{\_}{R}}_{11}-\frac{m_{11}}{}=a_{11}{\stackrel{\_}{R}}_{11}+a_{12}{\stackrel{\_}{R}}_{12}i{\stackrel{\_}{R}}_{12}-\frac{m_{12}}{}=a_{21}{\stackrel{\_}{R}}_{11}+a_{22}{\stackrel{\_}{R}}_{12},& \left(3.3\right)\end{array}$

[0123] where {overscore (R)}_ij() define the following integral Fourier transformation: ${\stackrel{\_}{R}}_{\mathrm{ij}}\left(\right)=\left[{\stackrel{\_}{R}}_{\mathrm{ij}}\left(\right)\right]=\frac{1}{}{}_0^{}{R}_{\mathrm{ij}}\left(\right)e^{-i}d.$

[0124] Then the spectral density S₁₁() of x₁(t) can be obtained as $\begin{array}{cc}{S}_{11}\left(\right)=\frac{1}{2}{}_{-}^{}{R}_{11}\left(\right)e^{-i}d=\mathrm{Re}\left[{\stackrel{\_}{R}}_{11}\left(\right)\right],& \left(3.4\right)\end{array}$

[0125] where Re denotes the real part.

[0126] Since R_ij()0 as , it can be shown that $\left(\frac{R_{\mathrm{ij}}\left(\right)}{}\right)=i{\stackrel{\_}{R}}_{\mathrm{ij}}\left(\right)-\frac{1}{}{R}_{\mathrm{ij}}\left(0\right)$

[0127] and equation (3.3) is obtained using this relation.

[0128] Solving equation (3.3) for {overscore (R)}_ij() and taking its real part, gives $\begin{array}{cc}{S}_{11}\left(\right)=\frac{-\left(a_{11}{m}_{11}+a_{12}{m}_{12}\right){}^2+A_2\left(a_{12}{m}_{12}-a_{22}{m}_{11}\right)}{\left[{}^4+\left(A_1^2-2A_2\right){}^2+A_2^2\right]},& \left(3.5\right)\end{array}$

[0129] where A₁a₁₁a₂₂, and A₂a₁₁a₂₂a₁₂a₂₁.

[0130] Expression (3.5) is the general expression for a narrow-band spectral density. The constants a_ij, i, j1,2, can be adjusted to obtain a best fit for a target spectrum. The task is to determine non-negative functions D₁²(x₁x₂) and D₂²(x₁x₂) for a given p(x₁,x₂).

[0131] Forming filters for simulation of non-Gaussian stochastic processes can be derived as follows. The Fokker-Planck-Kolmogorov (FPK) equation for the joint density p(x₁,x₂) of x₁(t) and x₂(t) in the stationary state is given as $\frac{}{x_1}\left(\left(a_{11}{x}_1+a_{12}{x}_2\right)p-\frac{1}{2}\frac{}{x_1}\left[D_1^2\left(x_1,x_2\right)p\right]\right)+\frac{}{x_2}\left(\left(a_{21}{x}_1+a_{22}{x}_2\right)p-\frac{1}{2}\frac{}{x_2}\left[D_2^2\left(x_1,x_2\right)p\right]\right)=0$

[0132] If such D₁²(x₁,x₂) and D₂²(x₁,x₂) functions can be found, then the equations of forming filters for the simulation in the Stratonovich form are given by $\begin{array}{cc}{\stackrel{.}{x}}_1=a_{11}{x}_1+a_{12}{x}_2-\frac{1}{4}\frac{}{x_1}{D}_1^2\left(x_1,x_2\right)+\frac{D_1\left(x_1,x_2\right)}{\sqrt{2}}{}_1\left(t\right),{\stackrel{.}{x}}_2=a_{21}{x}_1+a_{22}{x}_2-\frac{1}{4}\frac{}{x_2}{D}_2^2\left(x_1,x_2\right)+\frac{D_2\left(x_1,x_2\right)}{\sqrt{2}}{}_2\left(t\right),& \left(3.6\right)\end{array}$

[0133] where _i(t), i1,2, are two independent unit Gaussian white noises.

[0134] Filters (3.1) and (3.6) are non-linear filters for simulation of non-Gaussian random processes. Two typical examples are provided.

[0135] Example 1: Consider two independent uniformly distributed stochastic process x₁ and x₂, namely, $p\left(x_1,x_2\right)=\frac{1}{4{}_1{}_2},$

[0136] ₁x₁₁, ₂x₂₂.

[0137] In this case, from the FPK equation, one obtains $a_{11}-\frac{1}{2}\frac{{}^2}{x_1^2}{D}_1^2+a_{21}-\frac{1}{2}\frac{{}^2}{x_2^2}{D}_2^2=0,$

[0138] which is satisfied if

[0139] The two non-linear equations in (3.6) are now $\begin{array}{cc}{\stackrel{.}{x}}_1=\frac{1}{2}{a}_{11}{x}_1+a_{12}{x}_2+\sqrt{-\frac{a_{11}}{2}\left({}_1-x_1^2\right)}{}_1\left(t\right){\stackrel{.}{x}}_2=\frac{1}{2}{a}_{22}{x}_1+a_{21}{x}_2+\sqrt{-\frac{a_{22}}{2}\left({}_2-x_2^2\right)}{}_2\left(t\right),& \left(3.7\right)\end{array}$

[0140] which generate a uniformly distributed stochastic process x₁(t)with a spectral density given by (3.5).

[0141] Example 2: Consider a joint stationary probability density of x₁(t) and x₂(t) in the form

p(x₁,x₂)()C₁(b),b>0,>1, and $=\frac{1}{2}{x}_1^2-\frac{a_{12}}{2a_{21}}{x}_2^2.$

[0142] A large class of probability densities can be fitted in this form. In this case $D_1\left(x_1,x_2\right)=-\frac{2a_{11}}{-1}\left(+b\right),D_2\left(x_1,x_2\right)=\frac{2a_{11}{a}_{12}}{a_{21}\left(-1\right)}\left(+b\right)$ $\mathrm{and}$ $p\left(x_1\right)=C_1{}_{-}^{}{\left(\frac{1}{2}{x}_1^2-\frac{a_{12}}{2a_{21}}{u}^2+b\right)}^{}du.$

[0143] The forming filter equations (3.6) for this case can be described as following $\begin{array}{cc}\begin{array}{c}{\stackrel{.}{x}}_1=a_{11}{x}_1+a_{12}{x}_2-\frac{2a_{11}^2}{{\left(-1\right)}^2}\left[\frac{1}{2}{x}_1^2-\frac{a_{12}}{2a_{21}}{x}_2^2+b\right]x_1-\\ \frac{2a_{11}}{\sqrt{2}\left(-1\right)}\left[\frac{1}{2}{x}_1^2-\frac{a_{12}}{2a_{21}}{x}_2^2+b\right]{}_1\left(t\right)\end{array}\begin{array}{c}{\stackrel{.}{x}}_2=a_{21}{x}_1+a_{22}{x}_2+\frac{2a_{22}^2{a}_{12}^3}{{\left(-1\right)}^2}\left[\frac{1}{2}{x}_1^2-\frac{a_{12}}{2a_{21}}{x}_2^2+b\right]x_2+\frac{2a_{22}{a}_{12}}{\sqrt{2}\left(-1\right)}\\ \left[\frac{1}{2}{x}_1^2-\frac{a_{12}}{2a_{21}}{x}_2^2+b\right]{}_2\left(t\right)\end{array}& \left(3.8\right)\end{array}$

[0144] If _ik(x,t) are bounded functions (see Appendix I) and the functions F_i(x,t) satisfy the Lipshitz condition F(xxKxx, Kconst>0, then for every smoothly-varying realization of process y(t) the stochastic equations can be solved by the method of successive substitution which is convergent and defines smoothly-varying trajectories x(t). Thus, Markovian process x(t) has smoothly trajectories with the probability 1. This result can be used as a background in numerical stochastic simulation.

[0145] The stochastic differential equation for the variable x_i is given by $\begin{array}{cc}\frac{x_i}{t}=F_i\left(x\right)+G_i\left(x\right){}_i\left(t\right),i=1,2,,N,x=\left(x_1,x_2,,x_N\right).& \left(4.1\right)\end{array}$

[0146] These equations can be integrated using two different algorithms: Milshtein; and Heun methods. In the Milshtein method, the solution of stochastic differential equation (4.1) is computed by the means of the following recursive relations: $\begin{array}{cc}{x}_i\left(t+t\right)=\left[F_i\left(x\left(t\right)\right)+\frac{{}^2}{2}{G}_i\left(x\left(t\right)\right)\frac{G_i\left(x\left(t\right)\right)}{x_i}\right]t+G_i\left(x\left(t\right)\right)\sqrt{{}^{2}t}{}_i\left(t\right),& \left(4.2\right)\end{array}$

[0147] where _i(t) are independent Gaussian random variables and the variance is equal to 1.

[0148] The second term in equation (4.2) is included because equation (4.2) is interpreted in the Stratonovich sense. The order of numerical error in the Milshtein method is t . Therefore, small t(i.e., t110⁴ for ²1) is be used, while its computational effort per time step is relatively small. For large , where fluctuations are rapid and large, a longer integration period and small t is used. The Milshtein method quickly becomes impractical.

[0149] The Heun method is based on the second-order Runge-Kutta method, and integrates the stochastic equation by using the following recursive equation: $\begin{array}{cc}\begin{array}{c}{x}_i\left(t+t\right)=x_i\left(t\right)+\frac{t}{2}\left[F_i\left(x\left(t\right)\right)+F_i\left(y\left(t\right)\right)\right]+\\ \frac{\sqrt{{}^{2}t}}{2}{}_i\left(t\right)\left[G_i\left(x\left(t\right)\right)+G_i\left(y\left(t\right)\right)\right],\end{array}& \left(4.3\right)\end{array}$

[0150] where

[0151] The Heun method accepts larger t than the Milshtein method without a significant increase in computational effort per step. The Heun method is usually used for ²>2.

[0152] The time step t can be chosen by using a stability condition, and so that averaged magnitudes do not depend on t within statistical errors. For example, t510⁴ for ²1 and t110⁵ for ²15. The Gaussian random numbers for the simulation were generated by using the Box-Muller-Wiener algorithms or a fast numerical inversion method.

[0153] Table 5.1 summarizes the stochastic simulation of typical road signals.

TABLE 5.1
Types of	Types of		Results of
Correlation	Probability	Forming	Stochastic
Function	Density Function	Filter Equations	Simulations
	1D Gaussian
R y ( ) = 2 -	p ( y ) = 1 2 - 1 2 ( y - ) 2	y . + y = 2 ( t )
	1D Uniform
	p ( y ) = { 0 , y [ y 0 - y 0 + ] 1 2 , y [ y 0 - y 0 + ]	y . + 2 y = 2 2 ( 2 - y 2 ) ( t )
	1D Rayleigh
	p ( y ) = y 2 - ( y 2 2 2 )	y . + 2 y + 2 = 2 2 2 ( y + 2 ) ( t )
	2D Gaussian
R ( ) = 2 - [ cos + sin ]	p ( y 1 , y 2 ) = 1 2 1 2 1 2 ( ( y 1 - 1 1 ) 2 + ( y 2 - 2 2 ) 2 )	y + 2 y . + ( 2 + 2 ) y = 2 2 ( 2 + 2 ) ( t )
	2D Uniform
	p ( y 1 , y 2 ) = 1 4 1 2 - 1 < y 1 < 1 - 2 < y 2 < 2	y . 1 = 1 2 a 11 y 1 + a 12 y 2 + ( - a 11 2 ( 1 - y 1 2 ) ) 1 ( t ) y . 2 = 1 2 a 22 y 2 + a 21 y 1 + ( - a 22 2 ( 2 - y 2 2 ) ) 1 ( t )
	2D Hyperbolic
	p ( y 1 , y 2 ) = ( ) = C 1 ( + b ) - b > 0 ; > 1 = 1 2 y 1 2 - a 12 2 a 21 y 2 2	y . 1 = a 11 y 1 + a 12 y 2 2 a 11 2 ( - 1 ) 2 [ 1 2 y 1 2 a 12 2 a 21 y 2 2 + b ] y 1 2 a 11 2 ( - 1 ) [ 1 2 y 1 2 a 12 2 a 21 y 2 2 + b ] 1 ( t ) y . 2 = a 21 y 1 + a 22 y 2 2 a 22 2 a 12 3 a 21 3 ( - 1 ) 2 [ 1 2 y 1 2 a 12 2 a 21 y 2 2 + b ] y 2 + 2 a 22 a 12 2 a 21 ( - 1 ) [ 1 2 y 1 2 a 12 2 a 21 y 2 2 + b ] 2 ( t	FIG. 6F

[0154] FIG. 7 shows a vehicle body 710 with coordinates for describing position of the body 710 with respect to wheels 701-704 and suspension system. A global reference coordinate x_r, y_r, z_rr is assumed to be at the geometric center P_r of the vehicle body 710. The following are the transformation matrices to describe the local coordinates for the suspension and its components:

[0155] 2 is a local coordinate in which an origin is the center of gravity of the vehicle body 710;

[0156] 7 is a local coordinate in which an origin is the center of gravity of the suspension;

[0157] 10n is a local coordinate in which an origin is the center of gravity of the n'th arm;

[0158] 12n is a local coordinate in which an origin is the center of gravity of the n'th wheel;

[0159] 13n is a local coordinate in which an origin is a contact point of the n'th wheel relative to the road surface; and

[0160] 14 is a local coordinate in which an origin is a connection point of the stabilizer.

[0161] Note that in the development that follows, the wheels 702, 701, 704, and 703 are indexed using i, ii, iii, and iv, respectively.

[0162] As indicated, n is a coefficient indicating wheel positions such as i, ii, iii, and iv for left front, right front, left rear and right rear respectively. The local coordinate systems x₀, y₀, and z₀0 are expressed by using the following conversion matrix that moves the coordinate r along a vector (0,0,z₀) ${}_0^{r}T=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& 0\\ 0& 0& 1& z_0\\ 0& 0& 0& 1\end{array}\right]$

[0163] Rotating the vector r along y_r with an angle makes a local coordinate system x_0c, y_0c, z_0c0r with a transformation matrix _0c⁰T. $\begin{array}{cc}{}_{0c}^{0}T=\left[\begin{array}{cccc}\cos & 0& \sin & 0\\ 0& 1& 0& 0\\ -\sin & 0& \cos & 0\\ 0& 0& 0& 1\end{array}\right]& \left(5.1\right)\end{array}$

[0164] Transferring 0r through the vector (a_1n, 0,0) makes a local coordinate system x_0f, y_0f, z_0f0f with a transformation matrix ^0r_0fT. $\begin{array}{cc}{}_{0n}^{0c}T=\left[\begin{array}{cccc}1& 0& 0& a_{1n}\\ 0& 1& 0& 0\\ 0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]& \left(5.2\right)\end{array}$

[0165] The above procedure is repeated to create other local coordinate systems with the following transformation matrices. $\begin{array}{cc}{}_{1n}^{0n}T=\left[\begin{array}{cccc}1& 0& 0& a_{1n}\\ 0& \cos & -\sin & 0\\ 0& \sin & \cos & 0\\ 0& 0& 0& 1\end{array}\right]& \left(5.3\right)\\ {}_2^{1i}T=\left[\begin{array}{cccc}1& 0& 0& a_0\\ 0& 1& 0& b_0\\ 0& 0& 1& c_0\\ 0& 0& 0& 1\end{array}\right]& \left(5.4\right)\end{array}$

[0166] Coordinates for the wheels (index n: i for the left front, ii for the right front, etc.) are generated as follows. Transferring 1n through the vector (0, b_2n, 0) makes local coordinate system x_3n, y_3n, z_3n3n with transformation matrix ^1f_3nT. $\begin{array}{cc}{}_{3n}^{1n}T=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& b_{2n}\\ 0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]& \left(5.5\right)\\ {}_{4n}^{3n}T=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \cos {}_n& -\sin {}_n& 0\\ 0& \sin {}_n& \cos {}_n& 0\\ 0& 0& 0& 1\end{array}\right]& \left(5.6\right)\\ {}_{5n}^{4n}T=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& 0\\ 0& 0& 1& c_{1n}\\ 0& 0& 0& 1\end{array}\right]& \left(5.7\right)\\ {}_{6n}^{5n}T=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \cos {}_n& -\sin {}_n& 0\\ 0& \sin {}_n& \cos {}_n& 0\\ 0& 0& 0& 1\end{array}\right]& \left(5.8\right)\\ {}_{7n}^{6n}T=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& 0\\ 0& 0& 1& z_{6n}\\ 0& 0& 0& 1\end{array}\right]& \left(5.9\right)\\ {}_{8n}^{4n}T=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& 0\\ 0& 0& 1& c_{2n}\\ 0& 0& 0& 1\end{array}\right]& \left(5.10\right)\\ {}_{9n}^{8n}T=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \cos {}_n& -\sin {}_n& 0\\ 0& \sin {}_n& \cos {}_n& 0\\ 0& 0& 0& 1\end{array}\right]& \left(5.11\right)\\ {}_{10n}^{9n}T=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& e_{1n}\\ 0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]& \left(5.12\right)\\ {}_{11n}^{9n}T=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& e_{3n}\\ 0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]& \left(5.13\right)\\ {}_{12n}^{11n}T=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \cos {}_n& -\sin {}_n& 0\\ 0& \sin {}_n& \cos {}_n& 0\\ 0& 0& 0& 1\end{array}\right]& \left(5.14\right)\\ {}_{13n}^{12n}T=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& 0\\ 0& 0& 1& z_{12n}\\ 0& 0& 0& 1\end{array}\right]& \left(5.15\right)\\ {}_{14n}^{9n}T=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& e_{0n}\\ 0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]& \left(5.16\right)\end{array}$

[0167] Some of the matrices are sub-assembled to make the calculation simpler. $\begin{array}{cc}\begin{array}{c}{}_{1n}^{r}T={}_0^{r}T{}_{0n}^{0c}T{}_{1n}^{0n}T=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& 0\\ 0& 0& 1& z_0\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{cccc}\cos & 0& \sin & 0\\ 0& 1& 0& 0\\ -\sin & 0& \cos & 0\\ 0& 0& 0& 1\end{array}\right]\\ \left[\begin{array}{cccc}1& 0& 0& a_{1n}\\ 0& 1& 0& 0\\ 0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \cos & -\sin & 0\\ 0& \sin & \cos & 0\\ 0& 0& 0& 1\end{array}\right]\\ =\left[\begin{array}{cccc}\cos & 0& \sin & a_{1n}\cos \\ 0& 1& 0& 0\\ -\sin & 0& \cos & z_0-a_1\sin \\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \cos & -\sin & 0\\ 0& \sin & \cos & 0\\ 0& 0& 0& 1\end{array}\right]\\ =\left[\begin{array}{cccc}\cos & \sin \sin & \sin \cos & a_{1n}\cos \\ 0& \cos & -\sin & 0\\ -\sin & \cos \sin & \cos \cos & z_0-a_{1n}\sin \\ 0& 0& 0& 1\end{array}\right]\end{array}& \left(5.17\right)\\ \begin{array}{c}{}_{4n}^{r}T={}_{1n}^{r}T{}_{3n}^{1n}T{}_{4n}^{3n}T=\left[\begin{array}{cccc}\cos & \sin \sin & \sin \cos & a_{1n}\cos \\ 0& \cos & -\sin & 0\\ -\sin & \cos \sin & \cos \cos & z_0-a_{1n}\sin \\ 0& 0& 0& 1\end{array}\right]\\ \left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& b_{2n}\\ 0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \cos {}_n& -\sin {}_n& 0\\ 0& \sin {}_n& \cos {}_n& 0\\ 0& 0& 0& 1\end{array}\right]\\ =\left[\begin{array}{cccc}\cos & \sin \sin \left(+{}_n\right)& \sin \cos \left(+{}_n\right)& b_{2n}\sin \sin +a_{1n}\cos \\ 0& \cos \left(+{}_n\right)& -\sin \left(+{}_n\right)& b_{2n}\cos \\ -\sin & \cos \sin \left(+{}_n\right)& \cos \cos \left(+{}_n\right)& z_0-b_{2n}\cos \sin -a_{1n}\sin \\ 0& 0& 0& 1\end{array}\right]\end{array}& \left(5.18\right)\\ \begin{array}{c}{}_{7n}^{4n}T={}_{5n}^{4n}T{}_{6n}^{5n}T{}_{7n}^{6n}T=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& 0\\ 0& 0& 1& c_{1n}\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \cos {}_n& -\sin {}_n& 0\\ 0& \sin {}_n& \cos {}_n& 0\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& 0\\ 0& 0& 1& z_{6n}\\ 0& 0& 0& 1\end{array}\right]\\ =\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \cos {}_n& -\sin {}_n& 0\\ 0& \sin {}_n& \cos {}_n& c_{1n}\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& 0\\ 0& 0& 1& z_{6n}\\ 0& 0& 0& 1\end{array}\right]\\ =\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \cos {}_n& -\sin {}_n& -z_{6n}\sin {}_n\\ 0& \sin {}_n& \cos {}_n& c_{1n}+z_{6n}\cos {}_n\\ 0& 0& 0& 1\end{array}\right]\end{array}& \left(5.19\right)\end{array}$

[0168] Parts of the model are described both in local coordinate systems and in relations to the coordinate r or 1n referenced to the vehicle body 710.

[0169] In the local coordinate systems: $\begin{array}{cc}{P}_{\mathrm{body}}^2=P_{\mathrm{susp}.n}^{7n}=P_{\mathrm{arm}.n}^{10n}=P_{\mathrm{wheel}.n}^{12n}=P_{\mathrm{touchpoint}.n}^{13n}=P_{\mathrm{stab}.n}^{14n}=\left[\begin{array}{c}0\\ 0\\ 0\\ 1\end{array}\right]& \left(5.22\right)\end{array}$

[0170] In the global reference coordinate system r: $\begin{array}{cc}\begin{array}{c}{P}_{\mathrm{body}}^r={}_{1i}^{r}T{}_2{}^{1i}\mathrm{TP}_{\mathrm{body}}^2\\ =\left[\begin{array}{cccc}\cos & \sin \sin & \sin \cos & a_{1i}\cos \\ 0& \cos & -\sin & 0\\ -\sin & \cos \sin & \cos \cos & z_0-a_{1i}\sin \\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{cccc}1& 0& 0& a_0\\ 0& 1& 0& b_0\\ 0& 0& 1& c_0\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{c}0\\ 0\\ 0\\ 1\end{array}\right]\\ =\left[\begin{array}{c}{a}_0\cos +b_0\sin \sin +c_0\sin \cos +a_{1i}\cos \\ b_0\cos -c_0\sin \\ -a_0\sin +b_0\cos \sin +c_0\cos \cos -a_{1i}\sin \\ 1\end{array}\right]\end{array}& \left(5.23\right)\\ \begin{array}{c}{P}_{\mathrm{susp}.n}^r={}_{4n}^{r}T{}_{7n}{}^{4n}\mathrm{TP}_{\mathrm{susp}.n}^{7n}\\ =\left[\begin{array}{cccc}\cos & \sin \sin \left(+{}_n\right)& \sin \cos \left(+{}_n\right)& b_{2n}\sin \sin +a_{1n}\cos \\ 0& \cos \left(+{}_n\right)& -\sin \left(+{}_n\right)& b_{2n}\cos \\ -\sin & \cos \sin \left(+{}_n\right)& \cos \cos \left(+{}_n\right)& z_0+b_{2n}\cos \sin -a_{1n}\sin \\ 0& 0& 0& 1\end{array}\right]\\ \left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \cos {}_n& -\sin {}_n& -z_{6n}\sin {}_n\\ 0& \sin {}_n& \cos {}_n& c_{1n}+z_{6n}\cos {}_n\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{c}0\\ 0\\ 0\\ 1\end{array}\right]\\ =\left[\begin{array}{c}\left\{z_{6n}\cos \left(+{}_n+{}_n\right)+c_{1n}\cos \left(+{}_n\right)+b_{2n}\sin \right\}\sin +a_{1n}\cos \\ -z_{6n}\sin \left(+{}_n+{}_n\right)-c_{1n}\sin \left(+{}_n\right)+b_{2n}\cos \\ \left\{z_{6n}\cos \left(+{}_n+{}_n\right)+c_{1n}\cos \left(+{}_n\right)+b_{2n}\sin \right\}\cos -a_{1n}\sin \\ 1\end{array}\right]\end{array}& \left(5.24\right)\\ \begin{array}{c}{P}_{\mathrm{arm}.n}^r={}_{4n}^{r}T{}_{10n}{}^{4n}\mathrm{TP}_{\mathrm{arm}.n}^{10n}\\ =\left[\begin{array}{cccc}\cos & \sin \sin \left(+{}_n\right)& \sin \cos \left(+{}_n\right)& b_{2n}\sin \sin +a_{1n}\cos \\ 0& \cos \left(+{}_n\right)& -\sin \left(+{}_n\right)& b_{2n}\cos \\ -\sin & \cos \sin \left(+{}_n\right)& \cos \cos \left(+{}_n\right)& z_0+b_{2n}\cos \sin -a_{1n}\sin \\ 0& 0& 0& 1\end{array}\right]\\ \left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \cos {}_n& -\sin {}_n& e_{3n}\sin {}_n\\ 0& \sin {}_n& \cos {}_n& c_{2n}+e_{1n}\sin {}_n\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{c}0\\ 0\\ 0\\ 1\end{array}\right]\\ =\left[\begin{array}{c}\left\{e_{1n}\sin \left(+{}_n+{}_n\right)+c_{2n}\cos \left(+{}_n\right)+b_{2n}\sin \right\}\sin +a_{1n}\cos \\ e_{1n}\cos \left(+{}_n+{}_n\right)-c_{2n}\sin \left(+{}_n\right)+b_{2n}\cos \\ \left\{e_{1n}\sin \left(+{}_n+{}_n\right)+c_{2n}\cos \left(+{}_n\right)+b_{2n}\sin \right\}\cos -a_{1n}\sin \\ 1\end{array}\right]\end{array}& \left(5.25\right)\\ \begin{array}{c}{P}_{\mathrm{wheel}.n}^r={}_{4n}^{r}T{}_{12n}{}^{4n}\mathrm{TP}_{\mathrm{wheel}.n}^{12n}\\ =\left[\begin{array}{cccc}\cos & \sin \sin \left(+{}_n\right)& \sin \cos \left(+{}_n\right)& b_{2n}\sin \sin +a_{1n}\cos \\ 0& \cos \left(+{}_n\right)& -\sin \left(+{}_n\right)& b_{2n}\cos \\ -\sin & \cos \sin \left(+{}_n\right)& \cos \cos \left(+{}_n\right)& b_{2n}\cos \sin -a_{1n}\sin \\ 0& 0& 0& 1\end{array}\right]\\ \left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \cos \left({}_n+{}_n\right)& -\sin \left({}_n+{}_n\right)& e_{3n}\cos {}_n\\ 0& \sin \left({}_n+{}_n\right)& \cos \left({}_n+{}_n\right)& c_{2n}+e_{3n}\sin {}_n\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{c}0\\ 0\\ 0\\ 1\end{array}\right]\\ =\left[\begin{array}{c}\left\{e_{3n}\sin \left(+{}_n+{}_n\right)+c_{2n}\cos \left(+{}_n\right)+b_{2n}\sin \right\}\sin +a_{1n}\cos \\ e_{3n}\cos \left(+{}_n+{}_n\right)-c_{2n}\sin \left(+{}_n\right)+b_{2n}\cos \\ z_0+\left\{e_{3n}\sin \left(+{}_n+{}_n\right)+c_{2n}\cos \left(+{}_n\right)+b_{2n}\sin \right\}\cos -a_{1n}\sin \\ 1\end{array}\right]\end{array}& \left(5.26\right)\\ \begin{array}{c}{P}_{\mathrm{touchpoint}.n}^r={}_{4n}^{r}T{}_{12n}{}^{4n}T_{13n}^{12n}{\mathrm{TP}}_{\mathrm{touchpoint}.n}^{13n}\\ =\left[\begin{array}{cccc}\cos & \sin \sin \left(+{}_n\right)& \sin \cos \left(+{}_n\right)& b_{2n}\sin \sin +a_{1n}\cos \\ 0& \cos \left(+{}_n\right)& -\sin \left(+{}_n\right)& b_{2n}\cos \\ -\sin & \cos \sin \left(+{}_n\right)& \cos \cos \left(+{}_n\right)& z_0+b_{2n}\cos \sin -a_{1n}\sin \\ 0& 0& 0& 1\end{array}\right]\\ \left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \cos \left({}_n+{}_n\right)& -\sin \left({}_n+{}_n\right)& e_{3n}\cos {}_n\\ 0& \sin \left({}_n+{}_n\right)& \cos \left({}_n+{}_n\right)& c_{2n}+e_{3n}\sin {}_n\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& 1& 0& 0\\ 0& 0& 1& z_{12n}\\ 0& 0& 0& 1\end{array}\right]\left[\begin{array}{c}0\\ 0\\ 0\\ 1\end{array}\right]\\ =\left[\begin{array}{c}\left\{z_{12n}\cos +e_{3n}\sin \left(+{}_n+{}_n\right)+c_{2n}\cos \left(+{}_n\right)+b_{2n}\sin \right\}\sin +a_{1n}\cos \\ -z_{12n}\sin +e_{3n}\cos \left(+{}_n+{}_n\right)-c_{2n}\sin \left(+{}_n\right)+b_{2n}\cos \\ z_0+\left\{z_{12n}\cos +e_{3n}\sin \left(+{}_n+{}_n\right)+c_{2n}\cos \left(+{}_n\right)+b_{2n}\sin \right\}\cos -a_{1n}\sin \\ 1\end{array}\right]\end{array}& \left(5.27\right)\end{array}$