Method for Voltage Control of the Magnets of a Magnetically Levitated Railroad and Associated Control Unit

The invention is directed to a control method for regulation of the magnets of a magnetically levitated railroad, with the use of at least three variables, which are acquired in an observer unit (support circuit) based on the measured magnitudes for the magnet gap width s as well as the magnet acceleration b, as well as an associated control unit.

Such a method is known from the DE-Al 35 01 487. There a control loop for a magnetically levitated vehicle is described, which is guided along a track by controlled support- and guidance magnets. The control loop comprises an observer unit designated as a support circuit or support loop, which is fed by the magnet acceleration b in the direction of the magnetic attraction force, as well as the magnet gap width s by way of measured magnitudes. The support loop forms three variable magnitudes in the form of estimated values for the magnet gap width, the gap change velocity as well as the magnet acceleration with the help of summation or adding links, integrators and amplifier links. Theseestimated values are respectively fed to an amplifier or gain link, whose three output values are fed to an additional summation member, from which finally the control unit output signal can be derived. In general, such control loops are used in order to enable a stable levitation of the magnetic vehicle during standstill and a good following behavior at all traveling speeds. In the voltage control method described in the DE-Al 35 01 487 it is essentially intended to maintain the stability of the vehicle in a simple manner also when levitating at standstill. It is provided there for that reason to assign an adaptive rail observation unit to each support magnet, which is matched to the track vibrations, and which generates an adaptive signal, which is superimposed to the control loop of the same support magnet.

A control loop for an magnet elastically suspended at the levitating chassis--a magnetic wheel--has the form depicted in FIG. 1: herein the symbols mean:

v: traveling speed

b=z (z: magnet coordinate)

s=z--h (h: rail coordinate)

s.sub.o : required magnet gap

h and z are defined with respect to a fictitious inertial guide or pilot line.

The control unit utilizes signals which are measured directly at the location of the magnet, this being the object to be control. In order to assure the stability of a magnetic wheel in the course of voltage control, three variables have to be fed back. These are for instance the values s, s' and b=z. Since s' cannot be directly measured, this value is at least determined as an estimated value s from a reduced observer unit. Since the set of the three state variables thus obtained does not yet assure a sufficient following behavior, it is desirable to design the observer unit in such a way that it also supplies an approximate s.apprxeq.s.

The derivation of s or s from s and z is always connected with a differentiation of the rail coordinate. Thus, for example, s can be represented in Laplace presentation by an observer of the first order as ##EQU1##

Thus in limiting cases there applies: ##EQU2##

An improvement of the differentation (smaller .tau.) yields thus a high noise component because of high frequency rail disturbances, which must no longer be followed. This noise component can lead to instabilities because of the natural limits of the actuator. In case of a real guideway, which must be economical because of reasons of cost, there will thus always exist tolerances, which result in a high background noise component. From (1) one recognizes further that one does not have a following system or follow-up for the rail h.sub.(f), rather for the rail contour ##EQU3##

Even in case of a hard coupling to h there result thus changes in the gap because of the phase between h and h.

It follows from the above explanations that the rail tolerances permit only a minimum value of .tau. and thus a limited estimation of h, with this however they determine the optimum following or follow-up behaviour and with this the required air gap.

The invention is based upon the task to propose a method of the previously described type, with which the following or follow-up behavior of the magnetically levitated vehicle is improved, without increasing the background noise component. Furthermore, a control unit is to be designed which is suitable for performing this method.

This task is solved in the invention in that a rail signal is additionally fed to the controllers of the individual magnets (j), which is respectively obtained by a rail observer unit from the measured values a and b of one of the magnets (j--k) which is leading viewed in travel direction in such a way, that it represents a noiseless rail signal with a correct phase in the useful frequency range as referred to the respectively trailing magnets.

The first magnet of the vehicle must be controlled conventionally, however, the folllwing behavior can also be improved by a softer suspension to the chassis. It is also conceivable, that the signals s and b are determined upstream of the first magnet at a point of the levitating or floating chassis and that from this a rail signal for the first magnet is processed. This applies in principle also for the remaining magnets.

Methods are indicated in subclaims, which represent further expedient embodiment forms of the invention.

A control unit for performing the method in the invention will, to begin with, contain at least one control loop circuit arrangement for each magnet, which corresponds to the control loop (see there FIG. 1) described in the DE-Al 35 01 487 as far as its structure is concerned. Thus, the control loop circuit arrangement is to contain the following components: a first summation member for addition of the measured value b and of a first feedback signal, an integrator downstreamof the first summation member, a second summation member adding the output signal of the first integrator and a second feedback signal, a second integrator downstream of said summation member, as well as a third summation member subtracting the output signal of the second integrator from the measured value s, from the output signal of said summation member the two feedback signals are formed after multiplication of the factors depending on the cutoff frequency of the support loop. In accordance with the invention each magnet (j--k) is furthermore to be assigned at least one rail observer-circuit arrangement, whose construction is identical to that of the control loop circuit arrangement, whose feedback signals are formed by multiplication with factors, which are a function of the speed dependent cutoff frequency .omega..sub.v =2vD.sub.v /.DELTA.x and whose first feedback signal is feedable as a rail signal u to a trailing magnet (j).

Expedient embodiment forms of such a controller are described in additional subclaims.

In the following the invention will be described with particularity in the form of two embodiment examples with the help of drawings. It is shown schematically on:

FIG. 1 is a block diagram of a control loop.

FIG. 2 a state of the art control loop or circuit,

FIG. 3 a first controller unit in accordance with the invention,

FIG. 4 an additional controller unit in accordance with the invention,

FIGS. 5-10 the amplitude--and phase curves of different signals as a function of frequency.

The known support circuit in FIG. 2 contains a first summation member 1, which is fed by the measured value b as well as a first feedback signal, and whose output signal arrives at the input of a first integrator 3. The output signal of said integrator is supplied together with a second feedback signal to a second summation member 5, which is again wired upstream of a second integrator 4. The output signal of said second integrator reaches together with the measured value s a third summation member 2, where it is subtracted from the measured value s. The first or second feedback signals for the summation members 1 and 5 are derived from the output signal of the summation member 2 after multiplication by the factors .omega..sub.s.sup.2 or 2D.omega..sub.s in the amplification or gain members 6 or 7. The three estimated values s,s and s for the magnetic gap width as well as its first and second derivatives with respect to time are extraacted from the support loop as depicted in FIG. 2. .omega..sub.s represents the cutoff frequency and D the damping constant of the support loop.

The following expressions are calculated for the three estimated values from FIG. 2, wherein p represents the Laplace differential operator: ##EQU4##

It follows from this for low or high frequencies, respectively: ##STR1##

It is seen from equation (4) that an increase of the support loop cutoff frequency increases the background noise in the rail component.

FIG. 3 shows an embodiment form of the method in the invention or of a controller unit in the invention. To begin with a support loop 10 is shown for a magnet j trailing in traveling direction, which support corresponds to that in the FIG. 2. Thus the corresponding reference numbers have also been maintained. The estimated values s, s and s, which can be gathered from the support loop 10, are fed to the summation members 18, 9 and 8, which respectively receive an additional addend. The output signals of the summation members 8, 9 as well as 18 reach by way of amplification or gain members 11, 12, 13, where a multiplication of the input signals with the entered factors occurs, an additional summation member 14 which finally at its output yields the controller output signal u.sub.R.

Furthermore a rail observer is depicted in the FIG. 3, which pertains to a magnet j`k leading in traveling direction. The appropriate rail observer circuitry 20 equals that of the support circuit loop 10 as far as its structure is concerned. While, however, in the last named circuitry constant factors .omega..sup.2.sub.s as well as 2D.omega..sub.s are used in the amplification members 6 and 7, the factors in the corresponding amplification members 26 and 27 of the rail observer circuitry 20 are a function of the traveling speed v. The rail observer circuitry 20 contains as a matter of fact first, second and third summation members 21, 25 as well as 22 and two integrators 23, 24 analogous to the support circuit circuitry 10. The output signal of the amplification member 26 is fed as a rail signal u to the controller unit of the associated trailing magnet j. Herein additionally three amplification members 15, 16 and 17 are respectively wired-in as depicted in FIG. 3, in which the rail signal u is multiplied with the depicted factors for generation of the addends which have to be fed to the summation members 18, 9 as well as 8.

The rail signal u has the following form as can be easily calculated: ##EQU5##

Herein .omega..sub.v is given by: ##EQU6##

Herein .DELTA.x=x.sub.j-k.sup.-x j mean the distance between the magnets j--k and j, v the traveling speed, .tau..sub.k thus the time interval, by which the trailing magnet j lags the leading magnet j--k, as well as D.sub.v the damping constant of the rail observer.

Thus a factor .omega..sub.v depending on the speed is selected in the amplification members 26 and 27.

For the signal u.sub.j-k from a leading magnet j--k, which "sees" the rail sooner by the time interval .tau..sub.k compared to the magnet j the following holds: ##EQU7##

The e-factor expresses, that there exists a shift timewise between the rail signals h.sub.j--k at the location of the magnet j--k and h.sub.j at the location of the magnet j, said shift being a function of the traveling speed v. If now .omega..sub.s in the rail observer circuitry 20 of the magnets j--k is chosen in such a way that the following applies:

Then the rail signal u.sub.j--k for the magnet j in the range /p/<.omega..sub.v represents a non-delayed, however filtered signal -h for -h.

So that this remains valid for all speeds, the frequency .omega..sub.v must be changed corresponding to the traveling speed v.

A speed signal generated in a decentralized manner is advisable here for reasons of safety, such as it is proposed in the German Pat. No. 34 11 190 as well as the patent application P 35 15 350.4-32.

The rail portions in equation (2), measuring the estimated value h as well as its derivatives can be corrected with this signal h free of background noise. In the equation for s compared to s the following share is missing: ##EQU8##

If one substitutes h for h in this expression and adds it to h, then one gets ##EQU9##

A correction free of noise for h results from ##EQU10## and an improvement for h is given by ##EQU11##

In order to thus correct the estimated values s including the derivatives with respect to time according to equation (2), the expressions indicated above are added to the estimated values h including the derivatives for the rail coordinate with respect to time. From this it is seen that as depicted in FIG. 3 the rail signal u from the rail observer circuitry 20 is to be multiplied by the factor 1/N for correction of s, by the factor p/N for correction of s and by the factor 2D.omega..sub.s p/N for correction of s, where N=.omega..sup.2.sub.s +2D.omega..sub.s p+p.sup.2.

FIG. 4 represents an additional possibility to feed a rail signal u to the controller of the magnet j, which is obtained from the measured values s and b of a leading magnet j--k as viewed in traveling direction. The index v characterizes here again the leading magnet, the index s the trailing magnet which is to be controlled. The symbolism and the designations are selected to be analogous with the FIGS. 2 and 3. The switch 50 affords the possibility of switching from a position applying for standstill and traveling with very little background noise to a noise-free position for traveling (the left hand switch position).

A more detailed computational investigation of the controller unit in FIG. 4 shows that there the signals s and s are exactly equal to the output signals of the summation members 18 and 19 in the FIG. 3. The same applies for the output signals of the summation member 49 in FIG. 4 as well as of the summation member 8 in FIG. 3. As can be easily calculated with the help of the construction of the additional observer unit 80, there results as a correlation between the additional signal u and the rail signal u: ##EQU12##

In FIGS. 5-10 diagrams are depicted, in which frequency dependence of the amplitude--and phase behavior of the approximate--and estimated magnitudes h.sub.j, h*, h, h* as well as h compared to the real values h.sub.j and h is shown. On the abscissa the respective angular frequency .omega. is inserted on a logarithmic scale. The FIG. 5 shows the amplitude behavior of the auxiliary value h.sub.j for the parametric values v=111 m/s as well as .omega..sub.v =103,6 s.sup.-1. FIG. 6 shows the phase difference between these two magnitudes. It is seen that the phase lag amounts to a maximum of 10.degree. up to the cutoff frequency .omega..sub.v and that in the adjacent range the amplitude drops at the second order. FIG. 7 shows the amplitude behavior of the value h* improved in accordance with the invention compared to the estimated value h, both refer to the corresponding real magnitudes. It follows that the approximate magnitude h* improved in accordance with the invention approximates considerably better to the corresponding real value than the original estimated value h in equation (2). This applies according to FIG. 8 also for the phase difference between the two approximate--or estimated values h* and h on the one hand as well as the corresponding real values on the other hand. At least up to the cutoff frequency .omega..sub.v the approximate magnitude h* follows the corresponding real magnitude h as far as phase differences are concerned considerably better than the estimated magnitude h. Corresponding behavior can be derived from FIGS. 9 and 10 for the approximate magnitude h* in relation to the estimated magnitudes h. In all cases there thus results a considerably improved amplitude and phase behavior if approximate values given by the invention are utilized, which are formed from the rail signals of the rail observer circuitry of the respectively leading magnets.

The diagrams of the FIGS. 5-10 are based on the parametric magnitudes D=1, .omega..sub.s =10 s.sup.-1, .DELTA.x=1.5 m, D.sub.v =0.7, v=400 km/h as well as .omega..sub.v =103.6 s.sup.-1.

In the standstill state .omega..sub.v =0 that means the path of j--k is open toward k. During levitation at standstill all the advantages of the previous control concept apply. In addition, the invention which causes an improved following behavior, allows to further lower the value for .omega..sub.s, for instance to .omega..sub.s =5.sub.s.sup.-1, whereby the levitation stability while standing is increased and background noise is reduced.

The concepts estimated value, support circuit as well as observer unit used in this paper are common in control technology, especially in the case of magnetic levitation vehicles, as can be gathered from the following publications:

DE--A1 24 46 936,

DE--C2 31 17 971,

DE--A1 35 01 487,

"Einfuhrung in die Theorie der Beobachter" von J. Ackermann, Regelungstechnik, 1976, H. 7, S. 217-226 ["Introduction to the Theory of Observer Units" by J. Ackermann, Regulation Technology, 1976, Issue 7, pages 217-226];

"Regelung eines elektromagnetischen Schwebefahrzeuges mit integriertem Antriebs-, Trag- und Fuhrungssystem" von W. Vollstedt u. G. Kaupert, Regelungstechnik, 1978, H. 8, s. 258-265 ["Regulation of an Electromagnetic Levitating Vehicle with Integrated Drive-, Support- and Guidance System" by W. Volstedt and G. Kaupert, Regulation Technology, 1978, Issue 8, pages 258-265],

"Anwendung des magnetischen Rades in Hochgeschwindigkeitsmagnetschwebebahnen", von W. Gottzein, R. Meisinger u. L. Miller, ZEV-Glas. Ann. 103, 1979, Nr. 5, S. 227-232 ["Application of the Magnetic Wheel in High Speed Magnetic Levitation Trains", by W. Gottzein, R. Meisinger and L. Miller, ZEV-Glas. Ann 103, 1979, No. 5, pages 227-232].