CA1226330A - Method and apparatus for determining the flux vector of a rotating-field machine from the stator current and the stator voltage, and the application thereof - Google Patents

Abstract

ABSTRACT OF THE DISCLOSURE
An EMF detector supplies, from the voltage and the current, the EMF vector in a coordinate system fixed in space, the integral of which is the flux vector. The integration is carried out in the field oriented coordinate system, in which the EMF vector is transformed by means of a vector rotator. A
rotational component is taken into consideration in the integration by having a first integrator form, from the field parallel EMF component, the magnitude of the flux and having a divider form, from the second EMF component and the magnitude or the flux, the flux frequency, from which frequency an integrator forms simultaneously the flux angle and the angle of rotation for the vector rotator. An angle control input characterizing the steady state and the angle of rotation, through an angle servo control selectably furnishes the field frequency entered into the second integrator to form, from a magnitude servo control, a feedback signal added to the field parallel EMF component. Further provided are circuits for taking into account the stray inductance without differentiation and a circuit for stabilization and damping.

Description

33(;~

BACKGROUND OF THE INVENTION
The invention relates to a method for determining the flux vector of a rotating field machine from the stators current and the stators voltage and -the apparatus for implementing the method, as well as to an application of the method and apparatus.
Such a method is used with the apparatus described in German Offenlegungsschrift 30 26 202 for the field oriented operation of a rotating field machine fed by a (frequency) converter. For controlling the field orientation, the position of the flux vector is determined and the converter feeding the machine is controlled as a function of the position of -the flux vector in such a manner that the component of the stators current parallel to the flux and the stators current component perpendicular thereto can be influenced independently. Through the control of the stators current component parallel to the flux (magnetizing current), a preset value for the magnitude of the flux can be adjusted, while the current component perpendicular to the flux (active current) then enters into the torque linearly and can be used directly for the decoupled control of the speed of rotation or the torque.
However, knowledge of the position of the flux vector is necessary for this field orientation. In this connection it is advantageous to measure the flux, not directly via Hall probes, 33~

but to calculate it from electrical variables by means of a computer model circuit. The simplest possibility to accomplish this end is a so-called "voltage model" which determines the induced EM from -the input voltages of the motor by subtracting the ohmic stators voltage -lo-33~
1 I drop and the inductive leakage voltages. The flux it then

2 obtained as the integral of the EM.

3 For describing the machine current, machine if voltages, the EM and the flux plane vectors can be used each 5 I with two defining variables, for instance, their Cartesian or 6 polar components with respect to a stationary it stators 7 oriented coordinate system "fixed in space") or one co-8 rotating with the rotor shaft ("rotor oriented') or one 9 rotating with the field axis field oriented"). For the mentioned "voltage model", consideration in the stators 11 oriented Cartesian coordinate system is simplest because it is 12 merely necessary? for this purpose, to form, for instance, in 13 a three-phase machine, from the voltages and currents of the 14 three phases mutually shifted 120, by means of a "3/2"
coordinate converter, the corresponding Cartesian components 16 fixed in space such "stators oriented vector components are 17 characterized here by the subscript so and so) ox the 18 corresponding stators current vector is and the stators voltage '9 '! vector us. The vector en of the EM is then calculated, taking into consideration the stators resistance us and the 21 leakage inductance Jo through addition, component by 22 component, according to 23 t 24 I The twitter oriented Cartesian components of the flux vector 25 ¦¦ are then obtained a the integral of the corresponding 26 ¦¦ component of the EM vector. In a coordinate system rotating 27 I with the flux vector with the field-parallel coordinate axis 28 Jo and the field-perpendicular coordinate axis , the EM
29 I Hector has the "field oriented" components and AL, in ¦ the physical vector relation Skye I* , a rotational component _~_ ~2~336~
1 related to the flux frequency us (i.e., to the derivative ox 2 the angle pi between the axe and ) then appears in 3 accordance with TV I aft

4 Jo ,)<~

5 ' The voltage model is therefore always operated as

6 stators oriented.

7 The open integrators required for EM integration

8 have a tendency to drift and must be stabilized, for instance,

9 via a zero point control connected into a feedback line.
However, the correspondingly slow changes of the flux 11 component are alto suppressed at low operating frequencies 12 along with the zero drift. In addition, an angle (phase) error 13 is generated in stationary operation which has an effect, 14 likewise mainly at low frequencies, and leads to a disturbing mi.~orientation if the reference vows for the stators current 16 are given as field oriented. However, these disadvantages are 17 counterbalanced by the good dynamics of this voltage model.
to However, it is also possible to determine a model 19 l value for the machine flux from the machine currents (i.e., i the stators current vector us and in the case of a 21 synchronous machine, also the field current ) and the 22 j measured rotor position or from the rotor speed , which 23 frequently advantageous from a measurement point of view.
I;
24 Jo This "current model" simulates the processes occurring in the 25 ¦¦ machine electronically, a far as they lead to the development 26 111 of the flux. For this current model, the use Or a field 27 ¦ oriented coordinate system is of advantage, where the rotor 28 I time constant is taken into condonation a the time constant 29 I of a mouthing member and the current model forms a model flux ¦ frequent y, from which the flux angle can be wormed by 33~
1 integration.
2 The conversion from one coordinate system in o 3 another coordinate system rotated by a given angle is l accomplished by feeding the appropriate components of the 5 I vector to be transformed to a so-called "vector rotator", at 6 the phase input of which a corresponding phase signal, for 7 instance, sine and cosine of the angle of rotation, are 8 applied.
9 In the current model, model parameters as accurate as possible much be set for the machine parameters, so that, 11 for instance, temperature related changes of the rotor 12 resistance lead to falsifications of the model flux in sty to 13 as well as dynamic processes. For higher operating 14 frequencies, the voltage model is therefore to be preferred, by at low frequencies, the current model leads to a better 16 model value for the flux in spite of possible steady-state 17 inaccuracies.
18 In the mentioned German Of~enlegungsschrift 30 26 19 202.3, a combination of both models is therefore provided.
I According to the voltage model, there are formed prom the 21 I machine currents and the machine voltages two components of a 22 I model EM vector e So . from which the corresponding 23 components of the flux vector I related to this voltage 24 model are formed. The circuit operates here as stators lloriented and contains for the formation of the flux one 26 11 integrator for each Cartesian EM component. For stabilizing 27 these integrator, a component of this flux vector is 28 I impressed on a controller in a feedback line, the output 29 I' signal of which is impressed a a correction quantity for I correcting the corresponding component of the model EM vector I
1 on the integrator input. To the reference input of these 2 controller is fed the corresponding component of a model flux 3 vector formed by the stators currents and the rotor position 4 a the reference input ~'~.
The controllers therefore receive at their inputs 6 the Cartesian components, fixed in space, of the difference 7 vector sty s and furnish the Cartesian components, fixes 8 in space, of a correction vector; by impressing it on the 9 voltage model 9 the difference vector is leveled out on the average. Thereby, the voltage model, at least with respite to 11 its stationary behavior, is slaved to the current model, so 12 that the good dynamics of the voltage model is preserved, but 13 the better stationary flux determination ox the current model 14 is utilized at low frequencies.
The outputs of the two known correction controls 16 represent the stators oriented Cartesian components of a 17 correction vector, which rotates essentially with the 18 frequency of the vector US . The controllers must therefore 19 continuously process alternating quantities, which may be a disadvantage not only at high operating frequencies but, in 21 particular, this presents difficulties if the method is to be 22 implemented with a microprocessor.
23 It is therefore an object of toe present invention 24 to provide another way of determining the flux vector of a 25 if rotating field machine.
26 ¦' SUMMARY OF THE INVENTION
_ 27 Jo This problem is solved by the method of the present 28 " invention. The present invention also starts out from the 29 I concept that first, the EM vector ox the rotating field 30 I machine it formed by means of an EM detector from the voltage 33q;~

and the current, which is then modified by a feedback signal derived from the determined flux vector so as to form the flux vector by integration of the modified EM vector.
According to two basic embodiments, a rotating orthogonal coordinate system is taken as the basis for the modification of the EM vector, in which the EM components are processed further. Since ultimately, the one coordinate axis points in the direction of the flux vector, as will be explained later on, the transformed EM vector therefore has the orthogonal components eye and eye The flux vector is now formed by this modified EM through a special integration procedure.
Thus, in accordance with a broad aspect of -the invention, there is provided, in a method for determining the flux vector of a rotating field machine from the stators current and the stators voltage comprising the following steps:
a) forming the EM vector of the machine from measured values of the current and voltage;
b) modifying the EM vector by means of a feedback signal derived from the determined flux vector, and c) forming the flux vector by integration of the modified EM vector, the improvement comprising, the following steps:
d) defining the modified EM vector with first and second orthogonal EM components in a rotating coordinate system;
e) forming the magnitude of the flux vector by integration of the first rotating EM component; and Eye f) forming the feedback signal, from the quotient of the second rotating EM component and the magnitude of the flux vector, as said frequency, of the flux vector and forming an angle signal through integration of the frequency which angle signal at the same time fixes the angle of rotation of -the rotating coordinate system and is the angle signal determining the direction of the flux vector.
In accordance with another broad aspect of the invention there is provided, in a method for determining the flux vector of a rotating field machine from the stators current and the stutter voltage comprising the following steps:
a) forming the EMFvector of the machine from measured values of the current and voltage;
b) modifying the EM vector by means of a feedback signal derived from the determined flux vector; and c) forming the flux vector by integration of the modified EM vector, the improvement comprising the following steps:
d) defining the modified EM vector as first and second orthogonal components, the first component equal -to the sum of a first orthogonal EM component in a rotating coordinate system and a first feedback signal and the second component equal to a second orthogonal component in the rotating coordinate system;
e) determining the magnitude of the flux vector by into-gyration of the first modified EM component and forming the first feedback signal from the control deviation between said magnitude and a magnitude control input; and -pa-~6~3~

f) forming a second feedback signal prom the control deviation between a signal determining the angle of rotation of the rotating coordinate system, and an angle control signal for the direction of the flux vector, as the frequency of the flux vector and integrating said frequency to form an angle signal which determines the angle of rotation and the direction of which determines the direction of the flux vector.
Lo accordance with another broad aspect of the invention there is provided, an apparatus for determining the flux of a rotating field machine from voltage and current comprising a) an EM detector which forms from measured values of the current and voltage an EM vector;
b) a computing stage which forms a modified EYE' vector by means of a feedback signal derived from the flux vector; and c) an:integration.stage.which forms the flux vector by integration ox the modified EM vector, the improvement comprising:
d) the computing stage containing means for forming the orthogonal EM components in a rotating coordinate system;
e) the integration stage containing, for forming a first integrator signal, a first integrator to which the first orthogonal EM component of the flux vector in the rotating reference system or, if the coordinate axis is parallel to the flux, the magnitude coordinate of the vector; and f) an angle signal former with a second integrator for forming a second integrator signal, said second integrator acted upon by a feedback signal which is derived from a quantity of the determined flux vector and is related to the frequency of the flux -6b-~633~

vector, and said second integration signal being fed to the computing stage as the angle of rotation of the rotating coordinate system.
BRIEF DESCRIPTION OF THE DRAWINGS
Figure lo illustrates a operational amplifier with a multiplier in its negative feedback path.
Figure lb is a simplification of the circuit of Figure lay Figure to is a further simplification of what is illustrated in Figures land lb.
Figure 2 is a block diagram of a circuit for forming a signal which fixes a rotating orthogonal coordinate system and with which an EM vector determined by an EM detector ion a stators oriented coordinate system is transformed in-to a rotating coordinate system.
Figure 3 is a block diagram illustrating a simplify-cation of the system of Figure 2.
Figure 4 a block diagram illustrating a further simplification of Figure 3 utilizing the method of simplify-cation illustrated by Figures lay lb and to.
Figure 5 illustrates how the method illustrated by the -6c-~22~331~
1 block diagram of FIG. 4 can be built into the field oriented 2 control of a rotating yield machine fed by a frequency 3 converter.
4 FIG. 6 illustrates a modification of the arrangement of FIG. 5.
6 FIG. 7 illustrates further modification of the 7 system of FIG. 5.
8 FIG. 8 it a further alternate embodiment of the 9 arrangement ox the system of FIX. 5.
FIG. 9 is a diagram illustrating the locus of the 11 flux vector in a twitter oriented coordinate system for the 12 case where the rotating field machine operates at steady-state 13 but the integrators are subject to zero-point shift.
14 FIG. 10 is a diagram illustrating the aver gin of the component of the correction vector perpendicular to the do 16 component vector in one revolution along the locus curve.
17 FIG. 11 it a block diagram of a system in which a 18 correction vector is impressed on the vector of the 19 tran~ormed EM at the input of the circuit provided for the integration in the field oriented coordinate system.
21 FIG. 12 is a locus curve showing the angle between 22 the correction vector and the flux vector.

I To explain this method, a control loop with an operational amplifier 1 conceived as a very fat acting 26 controller it shown in FIG. lay the output variable 2 ox which 27 it impressed via a negative feedback loop with gain y on the 28 controller input along with an input x. If the gain of the 29 open integration amplifier 1 is designed as v, the mathematical structure shown next to it in FIG. lb is obtained ; -7- `
1. .

~.~2633~3 I for this circuit, to which applies 2 I I t I) 3 With sufficient gain, one obtains:

i.e., ultimately a division, as is also shown in FIG. to. FIG.
6 2 shows a circuit for forming the flux components no 7 z in a reverence system rotating with an (initially 8 arbitrary) frequency us , where the angle signal ye which 9 determines the respective instantaneous angle of rotation between the coordinates mixed in space (axes sly so) and the 11 rotating coordinate system ( I yo-yo ) is formed by a 12 corresponding frequency signal I, by means of an interrupter 13 2. By means of this angle I, , a vector rotator 3 forms, 4 from the stators oriented EM components JO, the transformed components L. It is particularly 16 advantageous to form from the angle pi by means of an angle 17 function generator 4 the angle functions coy I, and sin I:, 18 i.e., the stators oriented Cartesian components of a I corresponding unit vector us and to feed them to the vector rotator 3 as a corresponding pair of rotation angle signals, 21 so that the vector rotator 3 need still only execute a few 22 simple algebraic operations.
23 To arrive at the flux from the EM, it is now 24 basically not sufficient only to integrate the transformed component and pa (integrators 5 and 6); rather, the 26 ' rotational components y~s~.~l Casey' ~51 must I be fed to the~cintegrators in accordance with the already 28 ' mentioned vector relation. FIG. 2 shows the corresponding 29 feedback loops with multiplier 7 and 8.
3 The rotating coordinate system now becomes field .

~ZZ~3~
1 oriented if the coordinate axis I actually points in the 2 direction of the flux vector. Then, yo-yo and indicate 3 the field oriented EM components, and j and C
4 applies. This is achieved if a zero-point controller 9 which must operate very fast, sets in the frequency by 6 leveling out the component ye Thus, FIG. 2 indicates how 7 the EM vector must be integrated in the field oriented 8 coordinate system, taking the rotational components into 9 consideration.
Since, however, the fast controller 9 must always 11 force I as exactly as possible, the multiplier 7 in FIG.
12 2 can be omitted. This leads to the circuit shown in FIG. 3 13 in which a vector given by its Cartesian components is always 14 represented by a double arrow, in accordance with the two signal lines for its Cartesian components.
16 For the component taken over from FIG. 2, the save 17 reference symbols have been retained. It can now be seen from 18 this figure that with sufficiently high gain of the 19 subassembly 10 consisting of the integrator 6 and the very fast controller 9, the structure already discussed in FIG. 1 21 is generated, where the elements 6, 8 and 9 of FIG. 3 or the 22 elements 6 to 9 of FIG. 2 can be replaced by a single divider 23 11 (FIG. 4).
24 ; One thus arrives at a method, in which a signal (I: ) 1 is formed which fixes a rotating orthogonal coordinate 26 system (I I) and with which an EM vector determined 27 ¦, by an EM detector in a stators oriented coordinate system is 28 1 transformed into a rotating coordinate system (vector rotator 29 3). The field oriented components I, and L of the EM
I vector thus represent the modified EM vector-.
, ' , if ! I _ 9 _ 633~9 1 The absolute amount I I of the flux vector is now 2 formed by integration of the first modified EM compute 'y 3 (integrator 5). From the quotient I , the frequency 4 l; of the flux vector is wormed (divider 11). By integration of this frequency (integrator 2), the feedback signal I; is 6 obtainer, which determines at the same tire the angle of 7 rotation of the rotating coordinate system and the direction 8 of the flux vector. Thereby, the stators oriented polar 9 components of the Plus vector are already determined.
FIG. 5 now shows how this method can be build into 11 the Piled oriented control of a rotating field machine 21 fed 12 by a frequency converter 20, in this case an asynchronous 13 machine. By means of 3/2 coordinate converters 22 and 23 the 14 corresponding stators oriented vectors us and are formed from measured values for the current and the voltage, rum 16 which the EM detector 24 forms the stators oriented vector e-, 17 This is achieved in FIG. 5 by the provision that, by means 18 of multipliers 25 and 26 and a dif~erentiator 27 the vector I
19 i of the ohmic voltage drop and the vector RJ~ to of the leakage voltage drop are formed and are subtracted from the 21 voltage vector at the adding points 28 and 29.
I The vector rotator 3, the switching group 30 23 consisting of the elements 2, 5 and 11 and the angle function 24 generator 4 therefrom form the absolute amount of the flux 'I and the stators oriented Cartesian component of the unit 26 1, vector pointing in the direction of the flux vector. In z7 ' the field oriented coordinate ~y3tem, must be jet equal 28 to I and Lo in the qtator oriented coordinate I I septum the polar flux components ' and are 30 11 obtained by multiplication Or the unit vector components by I

, ., I i ~;z2~3~0 1 the amount of the flux of the stators oriented Cartesian components Jo and AL = sin I, of the vector 3 so I-Thus, the information required about the flux of the I rotating field machine for the field oriented operation of the 6 rotating field machine is available. In such a field oriented 7 operation? for instance, a speed control (speed control 31) 8 which controls the speed , i.e., the derivative of the angle g of rotation supplied by an angle-o~-rotation transmitter 32 to a corresponding reference value can be provided. The 11 output signal of the speed control 31 then supplies the 12 reference value or the torque to be supplied by the machine 13 for maintaining the speed or the active reference current ivy 14 proportional thereto. The "active current" is the torque forming component of the stators current which is perpendicular 16 to the flux. Toe control input for this active current can, 17 of course, also be predetermined by a torque control or 18 regulator or in some other manner.
19 In FIG. 5, a flux angle control 33 further shows that the flux of the rotating field machine can be controlled 21 to a predetermined reference value 1 I . The output signal 22 of this flux control 33 then makes available the control input I , for the "magnetizing current", i.e., the flux parallel 24 Jo component of the stators current. Frequently however, a flux I Al control is dispensed with and the magnetizing current is pro-26 1' jet in accordance with a constant flux in normal operation and 27 If a decreasing flux in the weakened field region. The control 28 ¦ input for the active current and the magnetizing current I I therefore represent the Cartesian components of the reverence 30 11 twitter current in the field oriented coordinate system vector I) 1, ,.

_ 1 1 _ 12 26 Jo 1 and it is only necessary to form, from these suitable field 2 oriented control inputs for the stators current which is 3 impressed on the machine via the frequency converter 2G and I its control unit 34, using the information regarding the flux 5 angle.
6 In the embodiment according to FIG. 5, the control 7 unit 34 has two separate inputs for the amount and the 8 direction (phase) of the stators current vector. The control g input for the amount of the current can be formed prom the field oriented reference vector Lo by means of a vector 11 analyzer 35 which makes available at its magnitude output the 12 reference magnitude lily and at its angle signal output the 13 pair of angle functions coy , sin for the reference I angle between the stators current vector and the axis (flux axis.
to This reference current given in the "field oriented"
17 coordinate system can be used as a reference value for current control, to which the corresponding field oriented components 19 of the actual current vector are then to be fed. To this end, Jo I a vector rotator 36 transforms the actual current vector 21 into the field oriented coordinate system by means of the flux 22 angle where a subsequent vector analyzer 37 furnishes the 23 i actual magnitudeli~ and the field oriented current angle 24 1 (angle between the flux and the current). my controlling the 25 I magnitude of the current determined in this manner (magnitude 26 !! control 38), the controlled quantity for the magnitude is 27 thereby formed for the control unit 34, while the 28 I corresponding angle control I protozoa the frequency control 29 I quantity in such a manner that the field oriented current I I angle become equal to the corresponding reference value Al`
1, If SLY
I For this yield oriented angle control the angle a , thy 2 tangent of the angle or another function of this angle can 3 basically be used in FIG. 5, the field oriented Cartesian , coordinate sin which is prevent at the vector outputs of I the vector analyzers 35 and 37 is used as the actual and the 6 I reverence value.
7 FIG. 5 is only an example ox a field oriented 8 control 9 where the control of the current to the reference g values preset in field orientation is performed in the field oriented coordinate system, i.e., the actual current value 11 are transformed by means of the vector rotator 36 into toe 12 field oriented coordinate system in order to be able to pro-set do quantities into the controllers 38 and 39. In this 4 case, the controller 39 can be relieved considerably by impressing on a subsequent adding point 40 the field frequency 16 in the sense of an anticipatory control.
17 Through the yield oriented control of the machine, 18 the control of the machine is ultimately determined by the 1g values determined according to the invention, for the flux magnitude 1~1 and the flux angle I,. At low frequencies, 21 however, the level of the measured voltage values required or 22 determining the flux it SO low that inaccuracies in the 23 I determination ox the flux can occur The inaccuracies are I ¦ noticeable primarily in ~teady-state operation with low I ¦¦ frequencies, while dynamic processes can still be determined 26 I by the device relatively accurately. For this reason, an 27 arrangement is shown in FIG. 6, in which the witching of 28 ¦ control input for the magnitude and angle of the flux is 29 Al controlled in such a way that stationary errors are largely I leveled out but the determination ox dynamic processes is If -13-r retained .
2 Since therefore, at low frequencies, the flux angle 3 is determined by a control input, in this case the unit vector ~;~
4 of a flux control vector , it is sufficient to readjust the ,1 angle or its unit vector accordingly by means of a 6 servo control 40. In FIG. 6, for determining the control 7 input ~,~ , a current model 41 fed by the actual current value and the rotor position angle is used and transformed by g means of a vector rotator 42 into the field oriented I coordinate system. This transportation corresponds to forming 11 an-angle difference. Instead of by the current model 41, the 12 control input can also be determined in another manner, for instance, from reference values formed in the field oriented control.
The angle servo control 40 therefore supplies the 16 frequency in such a manner that the direction of the field or vector determined corresponds in the average to the direction 18 given by Jo . The field frequency can therefore be taken off 19 at the output of the control 40, while the field direction tire., the one coordinate axis of tune field oriented 21 coordinate system) is formed by integration (integrator 43) 22 and subsequent information of the angular functions coy , 23 sin (function generator 44). It is shown in FIG. 6 that 24 I the direction information obtained in this manner can be used, lo for instance for the transformation of the stators voltage I I vector or other quantities from the twitter oriented coordinate 27 ' system into the field oriented coordinate system if the field I I coordinate control of the machine requires this (vector 29 ! rotator 36).
30 I Since the angle servo control 40 already furnishes !

~633C~
1 the field oriented direction of the flux vector, the elements 2 2 and 11, which in FIG. 5 are still necessary for determining 3 this direction, can be omitted in this circuit. However, this 4 1 leads to the situation that, thereby! the feedback signal derived from the output of the integrator 5 now comes out of 6 engagement and is therefore no longer needed. However, the 7 integrator 5 then still furnishes the integral of the field 8 parallel EM component, which is thematically equal to the g magnitude of the flux. This error, however, can be regulated out statically by feeding the integrator output to a 11 subtraction point 45 which supplies the control difference for 12 a magnitude servo control 46~ to which the magnitude control 13 input is fed as the reference value. The output signal of 14 this magnitude servo control 46 it impressed on the input of the integrator 5 together with the field oriented EM
16 component yo-yo .
17 Thus, according to FIG. 6, a circuit is arrived at, 18 in which the stators oriented EM vector furnished by the 19 EM detector 24 it modified such that by means of a signal Jo I determining rotating orthogonal coordinate system, the EM
21 vector 5 is transformed into the rotating coordinate system 22 vector rotator 3) and the feedback signal (output signal 23 of the magnitude servo control 46) is added to the yield 24 1 parallel component I of the transformed EM vector. The 25 I magnitude of the flux vector is then formed by integration of 26 it this doffed EM component, and the magnitude feedback signal 27 1 itself it determined from the control deviation of the I I magnitude and the magnitude control unit . From the 29 I control deviation of a signal determining the angle of Jo I rotation of the rotating coordinate system and of a if :~2~3~

control signal I for the direction of the flux vector, the frequency I of the flux vector and through integration of this frequency, the signal I is formed which determines at the same time the angle of rotation and the direction of the flux vector.
In the EM detector, the formation of the vector ~d~sldt leads to difficulties, because a mathematically exact differential lion of rapidly changing quantities can -technically not be carried out. In Figure 12 and Figure 13, of German Offend legungsschrift 30 34 275, however, a circuit is shown, in which;
through integration of a quantity a and subsequent subtraction of a quantity b with the integration constant -t permits calculating a quantity c which is given for -the case thank the untidy c is impressed on the integrator input with a negative sign, by C = (o, db/C~t)/(l + I
where the time behavior of a smoothing member with -the time con-slant t is designated by 1/(1 + sty).
By applying this circuit to the components of the vectors us and US , a circuit is arrived at according to Figure 7, which, by means of the integrator 50, the multipliers 51 and 52 addressed by -the stators resistance parameter us and the inductance parameter I as well as the subtraction points 53 and 54 allow calculating the vector s Jo s 5 - Ed So to which represents the vector of the EM of the machine, smoothed with the time constant t.

I

This smoothed EM vector e is formed by the --s assembly 55 and now has a time behavior which by itself would falsify the formation of the flux vector as the integral of -aye-33~
1 the EM vector. However, it this integration is carried out 2 component-wise by integrators 52 9 on the outputs of which are 3 impressed, at the additional point 57, the components of the I smoothed EM vector, multiplied by the lime constant t ox the smoothing (or the quotient of the time constant), then a 6 vector I dot is obtained, ire., the flux vector of 7 the rotating field machine. The integration time constant T
8 which must be introduced in all integrations for reasons of 9 normalization only has the effect of a proportionality factor, as everywhere in the circuits considered, which does not haze 11 to be discussed in detail here.
12 The integration in the assembly 5B can also be performed in the manner already described in connection with 14 Fits. 1 to 4 by transformation of the smoothed EM vector into the rotating reference system (vector rotator 3) and subsequent integration in this rotating reference system.
17 This it shown in FIG. 8. The angle of rotation required for 18 the transformation at the angle signal input of the vector 19 rotator 3 is formed from the integral of the feedback signal which is the quotient of the divider 11, having as inputs the I second component and the integral first component, integrated 22 in integrator 5, of the modified EM vector taken off at the 23 I terminal 60. Since the element 2 to 5 and 11 form the 24 I integral of the vector 5 the output signal of the 25 Al integrator 5 is the magnitude ox the smoothed flux vector Jo 26 , belonging to the smoothed EM. Similarly, the integrator 2 or 27 Al the function generator 4 furnishes the direction angle I, or 28 if the corresponding unit vector of the smoothed flux vector 29 I which, for dynamic processes, deviates prom the direction I, of the actual (unsoothed) flux vector due to the smoothing I!

633~
1 effect of the assembly I Therefore, the first rotting EM
2 component yo-yo. applied to the terminal 60 corresponds to the 3 component perpendicular to the smoothed flux, of the smoothed 4 ! EM, while the second component corresponds to the corresponding component, perpendicular to the smoothed flux 6 vector ~'~, ox the smoothed EM.
7 The effect ox the smoothing can now be compensated 8 by means of the multiplier 61 and the adding stage 62 in that 9 the quantity which is supplied by the integrator 5 a is, to as the magnitude of the smoothed flux vector, equal to the 11 component of this vector, parallel to the smoothed flux 12 vector, it added in the amount r ' to i, i.e., a signal 13 proportional to the smoothing time constant t and the first 14 component ( I ) of the modified EM vector. Thereby, the first component I, a the flux vector is formed in the 16 coordinate system I" . Since the second component of 17 the smoothed flux vector in this coordinate system is zero, 18 the second component of the smoothed flux vector can be taken 19 off directly at the output of the multiplier 63 as a signal which is proportional to the smoothing tire constant and the 21 second component AL of the modified EM vector.
22 Due to the direction difference between the smoothed 23 flux vector (angle of rotation I) and the actual flux angle (I
24 1 there are therefore available at the output of the assembly I Jo 58 the Cartesian components of the actual flux vector, but in 26 Al a coordinate ~y3tem oriented to the smoothed flux vector. A
27 vector analyzer 64 can determine therefrom the actual flux 28 I magnitude as well a the angle difference between the 29 11 actual flux vector and the smoothed flux vector, i.e., the angle or it Cartesian components coy and sin ~633~
1 oriented to the smoothed flux angle. The actual twitter 2 oriented flux direction I is then obtained as the sum of the 3 angles (vector rotator 65).
4 If the frequency of the actual flux vector is required, it must be determined from the frequency I, of the 6 smoothed flux vector and the derivative of the angle 7 The frequency corresponding to this derivative, however, 8 rarely amounts to more than 0.1~ of the smoothed flux g frequency. If this frequency is needed, for instance, only for an anticipatory control of an angle control according to 11 FIX. 5 or for the damping circuit explained in the following, 12 this supplemental frequency can be ignored, so that the frequency of the smoothed flux vector can be used as the flux 4 frequency in good approximation.
Integrators in general have a tendency for their 16 zero point to drift away and of other integration errors which 17 are noted as disturbing especially at low frequencies 18 stationary and quasi-stationary states ox the rotating field 19 machine). In FIG. 9, the locus of the flux vector for the case that the rotating field machine operates steady-state, 21 but the integrators used are subject to a zero-point shift is 22 given in a stators oriented coordinate system sly so. The 23 locus curve of the determined flux vector is then eccentric, I Al i.e., the center 0 of the locus curve is shifted relative to 25 I the coordinate origin Ox by a vector , the "d-c component"
26 I of the flux vector. The vector components numerically 27 1' determined by the circuit are then mixed quantities, in which 28 l: the sinusoidal motion of the twitter oriented Cartesian 29 I coordinates of the actual machine flux lo superimposed on the 3 ¦ respective Cartesian component of the d-c component factor a .
i _19_ , 1, 33~

In order to suppress this d-c component in the steady-state case, a correction vector which is derived from the flux vector, is advantageously added to the EM vector in such a manner that it becomes zero for uniform rotation of the flux vector US and suppresses during steady-state operation of the rotating field machine a d-c- component in the locus curve of the flux vector fixed in space. The magnitude of the correction vector should therefore be proportional to a "transient quantity" of the flux vector, where a transient quantity is understood to be a mahogany-tune of the flux vector which disappears if the locus curve is centered (uniform rotation).
Such a transient quantity is, for instance, -the angular v O
acceleration US of -the flux vector which is positive for positive rotation above the line Ox - O and negative in the half-plane below Co - C . Another preferred transient quantity is the derivative with respect to time of the magnitude of the flux, where the sign is the reverse.
If the correction vector is now given a direction which is perpendicular to the direction of the flux itself, then one has for instance, in the field oriented coordinate system:

O, d` a Stators oriented, the picture shown in Figure 10 is then obtained, in which the correction vector is obtained from the flux vector us by a rotation by I according to the negative sign of , i.e., corresponding to the positive sign of . When going through the straight line OX - O , one has I= Therefore the correction vector becomes zero for this crossing, and -then is rotated relative to the vector sty the Sue 1 angle Jo AL in the Hal plane at the right, bottom, 2 according to the sign reversal by the angle I , relative 3 to the vector I.
4 . As how in FIG. 10, the component of the correction I vector perpendicular to the d-c component vector G is 6 averaged out in one revolution along the locus curve, Chile 7 the component parallel to tune d-c component vector always 8 has a direction opposed to the d-c component vector.
9 Impressing this correction vector on the vector of the transformed EM at the input of the circuit provided for 11 the integration in the field oriented coordinate system 12 therefore has the effect that the locus curve of the flux 13 vector obtained by the integration is shifted the more in the 14 direction opposed to the d-c component vector, the larger this d-c component vector is. In the steady-state, however, the 16 correction vector disappears completely and therefore does not 17 lead to a steady state error averaging of the flux. Such a 18 circuit is shown in FIG. 11.
19 In this circuit new elements include a magnitude servo control 71 as well as a double-throw switch 72. If the 21 two controllers 70 and 71 are put out of engagement by closing 22 their ghort-circuit switches, and the double-throw switch 72 23 is put in the position shown, the configuration explained in I 11 FIG. 8 will result. The EM detector 55, which forms, prom ¦ measured values of current and voltage, the EM vector or the 26 if smoothed vector us , is preceded by a d-c component control 27 I which determines with relatively low gain the d-c components 28 If in the components of the voltage vector and subtracts them, I 1 component by component from the voltage vector. This d-c 1 component control 73 is designed to be so weak that it causes I , 3~C~
1 practically no phase distortion of the voltage vector.
The EM detector is followed by a computing stage 74 3 which supplies the modified EM vector and contains the vector j rotator 3 which forms the orthogonal EM components in a I rotating coordinate system which is rotated relative to the 6 stoker oriented coordinate system by an angle of rotation I, .
7 To the computing stage 74 is connected the integration stave 8 58, the integrator 5 of which supplies the amount of the flux 9 vector (in this case, initially the magnitude of the smoothed flux vector). The angle of rotation ye in turn is supplied by 11 an angle signal wormer which contains the second integrator 2 12 as well as also the function generator 4 if pairs of angle 13 function are always used as angle signals. The input signal 14 of this integrator is the signal I, fed back from the integration stage 58, which indicates the frequency of the 16 smoothed flux vector.
17 The correction vector explained above is supplied 18 by a correction vector generator. In the simple case 19 considered Jo far, where is perpendicular to the flux vector or the smoothed flux vector and therefore contains only 21 one component or z perpendicular to the flux, the 22 correction vector generator needs to contain only one signal 23 line with a differentiator 76. branching off at the magnitude 24 , output of the vector analyzer 64. Also this differentiation 25 11 is generally superfluous, since the derivative of the flux 26 I magnitude e~9entially agrees with the input signal of the 27 11 integrator 5 and therefore, the magnitude of the correction 28 Al vector can be taken off with sufficient accuracy at the 29 It corresponding component of the transformed EM vector or the 30 I modified EM vector 9 respectively.
.1 ,.
, -22-If l!

isle 1 Through the correction vector, not only is a d-c 2 component control ultimately achieved, but also a damping ox 3 , the entire flux determining device. This, however, degrades j the dynamics of the flux determination. However, such degradation can be prevented if the magnitude of the 6 correction vector is not determined by I' alone, but from the 7 difference 'I where is a control input for the change 8 of the flux. In particular, cay be taken off from the 9 current model mentioned at the outset or from the reference values of the rotating field machine control.
11 It was found further that it it advantageous to 12 preset the correction vector not always exactly perpendicular 13 to the flux vector or the smoothed flux vector. Especially at 14 low frequencies it is advantageous if the correction vector also has a component parallel to the flux vector. In FIG. to, 16 the field oriented locus curve of an advantageously preset 17 control vector for the individual given values of the flux 18 frequency us or it given.
19 It is seen therefrom that at frequencies above Owl, the component perpendicular to the field substantially 21 predominates over the field parallel component. While at low 22 frequencies the angle of approximately 90 is shifted to 23 i about _ 180, the value which belongs to the frequency 24 zero and for which the component perpendicular to the field I would disappear completely, is not reached since, here, a 26 I Singular operating state is involved in which the flux vector 27 Jo no longer rotates but is at rest. If the device is also to 28 I serve for determining the flux in the rest condition, 29 I impure sing ox the correction value is disabled.
pa I Entering the angle Jo between the correction vector .., Jo ~2633~
1 and the flux vector by the programmed locus curved according 2 to FIG. 12, means that the correction vector in the rotating 3 coordinate system is based on the control vector , the 4 magnitude of which is modified by multiplication with the transient quantity. The correction vector formed thereby thus 6 not only has an angle depending on or 9 but its 7 magnitude is proportional to the transient quantity via a 8 proportionality factor likewise dependent on (namely ! the 9 function-dependent given magnitude of the control vector).
If the device according to FIG. 11 is used to 11 intervene with the determined flux vector into the control of 12 the rotating field machine, it may turn out to be advantageous 13 to also change this locus curve of the control vector in 14 dependence on the operation. In particular, it may be advantageous to change the control vector as a function of 16 the load condition, for instance, the angle between the to current and the voltage of the rotating field machine, or of 18 another state variable W of the machine.
19 The engagement via the state variable W can optionally act 90 that the correction vector, after being 21 averaged over one revolution of the flux vector on the locus 22 curve is no longer anti parallel to the d-c component vector 23 and this d-c component vector is therefore not regulated out 24 I correctly, buy that nevertheless a stable steady-state " operation of the machine is achieved by the interaction with 26 , the machine and it control 27 The negative branch ( O ) in the locus curve 28 1. according to FIG. 12 relates to the cave that the flux 29 ' revolve in a mathematically negative sense. In this case, , the quantity is positive in the left-top half-plane in the j,. ..
,, '_ l ~24-,, i33~

eccentric locus curve as per figure 10 and negative in the other half-plane, so that here also the same direction ox the control vector is obtained in both half planes The setting of the control vector Jo and respectively, corresponding to Figure 12 can therefore be accomplished in the device according to Figure 11 by a function memory 75 (PROM) which is addressed by the input signal us of the second integrator 2 and, if applicable, by the load angle of the rotating field machine or another operating quantity W
of the rotating field machine. This control vector can then be multiplied component by component by -I I or by a difference formed by means of the reference input I so that it can subsequently be fed to a summing point 77 in the computing circuit 74. In some circumstances, depending on whether operating as a motor or a generator, angles f q~ are also possible.
As has been explained already, the outputs of the integrating stage 58 represent the Cartesian components of the unsoothed flux vector oriented to the smoothed flux vector.
Conversion of the Cartesian components into polar components or conversion into other coordinate systems presents no difficulties. The vector analyzer 64, for instance determines the magnitude coordinate and the angle coordinate my Processed as a unit vector). The vector rotator 65 therefrom forms the stators oriented angle coordinate ~5-~5'~ , and a multiplier 80 can supply from this angle coordinate shown as a unit vector and the 3;3 magnitude coordinate the stators oriented Cartesian coordinates of the vector I O It is further shown in Figure 11 that an actual vector or a reference vector (for instance, in a rotating--aye-~6~3~
I field machine control according to FIG. 6, the stators current 2 vector) can be transformed by mean of transformation by 3 j the vector rotator 81 driven by I into the - coordinate system and by means of a vector rotator 82 into the stators Oriented coordinate system, and is then processed further 6 j later on in the rotating field machine control.
7 If now the double-throw switch 72 is switched from 8 the position shown into the other position, the configuration 9 of FIG. 6 is obtained, in which the divider 11 is no longer engaged. While the integrator 2, which, however, is now 11 preceded by the angle servo control 719 still acts as an angle 12 wormer for forming the angle of rotation , it levels out, 13 if it is activated by opening its switch, the angle of 14 rotation to the reference angle (which may be g Yen, for instance, by the current model), by changing the frequency 16 for the angle ox rotation us formed by the integrator 2.
17 Since in this arrangement, the angles are used as pairs of 18 angle signals, the angle difference is formed by a vector 19 rotator 83, of which, however, only one component is needed for addressing the angle servo control 71, for instance, the 21 component Sweeney I 1 As already explained, in this case the amplitude 23 I servo control 70 is also activated by opening its short 24 Circuit switch. This control controls the difference YO-YO
fin accordance with an amplitude control input and the 26 ! amount of the Plus taken opt at the output of the 27 integrator 5 by impressing its control output signal on the 28 if component ~-~ of the modified EM vector.
29 1 In this switch position, a smoothing device for the ¦ EM provided according to FIG. 11 and the subsequent . , i3~6~
correction in the integration stave 58 continue to be in 2 engagement. Likewise, it this is desirable, the depone 3 provided by the correction vector can also remain switched on.
4 This switch position is provided particularly at low frequencies in the rotating yield machine. Also, the control 6 inputs determine substantially the determination of the flux.
7 Errors dye to the low level of the measured voltage values are 8 not noticeable in the steady state case while, however, the 9 good dynamics of the voltage model practically remains intact.
The device according to the present invention can be 11 used independently of the control of a rotating field machine 12 to monitor the flux of the rotating field machine for adjustment and control purposes. It can also be used to to intervene in the upper frequency range into the control of the machine, but in the lower speed range to idle along in slant-16 by operation if the machine it then controlled by other means 17 (for instance, by means of a current model). By suitably 18 setting in the reference input A it can be achieved 19 that in the lower speed range, in which the low voltage level of the voltage values makes the use of the voltage model more 21 difficult per so, the voltage model is determined in the 22 steady-state by the control inputs (for instance, a current 23 model), while the voltage model measures dynamic deviations 24 from the steady-state dynamically correctly. The transition , from the state controlled by the control inputs to the I uncontrolled state can be achieved discontinuously by simply switching the switch 72 into the state shown in FIG. if;
I however, a continuous transition is also possible, in which I the switch 72 is switched alternatively with a frequency-dependent duty cycle.

Claims (29)

WHAT IS CLAIMED IS:

1. In a method for determining the flux vector of a rotating field machine from the stator current and the stator voltage comprising the following steps:
a) forming the EMF vector of the machine from measured values of the current and voltage;
b) modifying the EMF vector by means of a feedback signal derived from the determined flux vector; and c) forming the flux vector by integration of the modified EMF vector, the improvement comprising, the following steps:
d) defining the modified EMF vector with first and second orthogonal EMF components in a rotating coordinate system;
e) forming the magnitude of the flux vector by integration of the first rotating EMF component; and f) forming the feedback signal, from the quotient of the second rotating EMF component and the magnitude of the flux vector, at said frequency, of the flux vector and forming an angle signal through integration of the frequency which angle signal at the same time fixes the angle of rotation of the rotating coordinate system and is the angle signal determining the direction of the flux vector.

2. In a method for determining the flux vector of a rotating field machine from the stator current and the stator voltage comprising the following steps:
a) forming the EMF vector of the machine from measured values of the current and voltage;
b) modifying the EMF vector by means of a feedback signal derived from the determined flux vector; and c) forming the flux vector by integration of the modified EMF vector, the improvement comprising the following steps:
d) defining the modified EMF vector as first and second orthogonal components, the first component equal to the sum of a first orthogonal EMF component in a rotating coordinate system and a first feedback signal and the second component equal to a second orthogonal component in the rotating coordinate system;
e) determining the magnitude of the flux vector by integration of the first modified EMF component and forming the first feedback signal from the control deviation between said magnitude and a magnitude control input; and f) forming a second feedback signal, from the control deviation between a signal determining the angle of rotation of the rotating coordinate system, and an angle control signal for the direction of the flux vector, as the frequency of the flux vector and integrating said frequency to form an angle signal which determines the angle of rotation and the direction of which determines the direction of the flux vector.

3. The method according to claim 1, comprising forming a vector from the measured values of current and voltage which corresonds to a smooth EMF, and transforming the orthogonal EMF components in the rotating coordinate system;
integrating the first component of the modified EMF vector;
forming a signal proportional to the smoothing time constant times the first component of the modified EMF vector; adding said integrated first component and said signal proportional the first Cartesian component; and forming a signal proportional to the smoothing time constant times the second Cartesian component of the modified EMF vector to form the second Cartesian component of the flux vector in the rotating coordinate system.

4. The method according to claim 1, comprising, deriving a correction vector from the flux vector which, in steady state operation of the rotating field machine, suprresses a d-c component in the locus curve of the flux vector fixed in space; and adding said correction vector to the transformed EMF vector.

5. The method according to claim 4, and further comprising setting in to the correction vector a predetermined direction difference relative to the flux vector, and a magnitude which is proportional to a magnitude of the flux vector which disappears when the locus curve is centered.

6. The method according to claim 5, wherein the correction vector is set in in such a manner that its direction is opposed, averaged over one revolution on the locus curve, to the d-c component in the locus curve fixed in space.

7. The method according to claim 5, wherein the magnitude of the correction vector is proportional to a predetermined proportionality factor determined at a function of the frequency of the flux vector and/or a state variable of the rotating field machine, proportional to the derivative of the flux magnitude or proportional to the first orthogonal EMF
component of the transformed EMF vector or proportional to the difference of the derivative, or to the first component of the difference of the derivative, or to the first component of the transformed EMF vector and a non-stationary control input.

8. The method according to claim 5, and further comprising adding to the direction of the correction vector component perpendicular to the flux vector which depends on the frequency of the flux vector and a state variable of the rotating field machine and does not vanish for frequencies other than zero.

9. The method according to claim 8, wherein the component perpendicular to the flux is positive or negative, depending on the sign of the product of the flux frequency and the magnitude of the correction vector.

10. The method according to claim 1, comprising forming a vector from the measured values of current and voltage a vector which corresponds to a smooth EMF;
transforming said vector to form the orthogonal EMF components in the rotating coordinate system; integrating the first component of the modified EMF vector; forming a signal proportional to the first component of the modified EMF vector times a smoothing time constant, adding said integrated first component and said signal proportional to form the first Cartesian component, and forming, from a signal proportional to the second component of the modified EMF vector, the second Cartesian component of the flux vector in the rotating coordinate system.

11. The method according to claims 10, and further comprisng deriving a correction vector from the flux vector which suppresses a d-c component in the locus curve of the flux vector fixed in space, in steady state operation of the rotating field macine and adding said correction vector to the transformed EMF vector.

12. The method according to claim 11, comprising giving to the correction rector, a predetermined direction difference relative to the flux vector and a magnitude which is proportional to a magnitude of the flux vector which is proportional to a quantity of the flux vector which vanishes if the locus curve is centered.

13. The method according to claim 12, comprising presetting the correction vector in such a manner that, averaged over one revolution on the locus curve, its direction is opposed to the d-c component in the locus curve fixed in space.

14. The method according to claim 12, wherein the magnitude of the correction vector is proportional to a predetermined proportionality factor determined as a junction of the frequency of the flux vector and/or a state variable of the rotating field machine proportional to the derivative of the flux magnitude or proportional to the first orthogonal EMF
component of the transformed EMF vector or proportional to the difference of the derivative or to the first component of the transformed EMF vector and to a non-stationary control input.

15. The method according to claim 12, comprising presetting a component for the direction of the correction vector, which is perpendicular and which depends on the frequency of the flux vector and in particular on the frequecy of the flux vector and a state variable of the rotating field machine and does not vanish for frequencies different from zero.

16. The method according to claim 15, wherein the component perpendicular to the flux is positive or negative depending on the sign of the product of the flux frequency and the magnitude of the correction vector.

17. The method according to claim 10 or claim 13, wherein the control inputs are formed from reference values of the machine or simulation of the flux from actual values for the current and the rotor position of the machine.

18. In apparatus for determining the flux of a rotating field machine from voltage and current comprising:
a) an EMF detector which forms from measured values of the current and voltage an EMF vector;
b) a computing stage which forms a modified EMF
vector by means of a feedback signal derived form the flux vector; and c) an integration stage which forms the flux vector by integration of the modified EMF vector, the improvement comprising:
d) the computing stage containing means for forming the orthogonal EMF components in a rotating coordinate system;
e) the integration stage containing, for forming a first integrator signal, a first integrator to which the first orthogonal EMF component of the flux vector in the rotating reference system or, if the coordinate axis is parallel to the flux, the magnitude coordinate of the vector; and f) an angle signal former with a second integrator for forming a second integrator signal, said second integrator acted upon by a feedback signal which is derived from a quantity of the determined flux vector and is related to the frequency of the flux vector, and said second integration signal being fed to the computing stage as the angle of rotation of the rotating coordinate system.

19. Apparatus according to claim 18, wherein said integration stage contains a divider for forming the quotient from the second orthogonal EMF component and the first integration signal, and wherein the output signal of the divider is fed to the input of the second integrator and the second integration signal to an output device for the angle or the Cartesian components of the flux vector.

20. Apparatus according to claim 18, and further including means to form a correction vector for setting in a correction vector which is rotated relative to the flux vector and, in particular, is not parallel to the flux vector, and disappears in the steady state, and means for summing the EMF

vector and the correction vector.

21. Apparatus according to claim 20, comprising means for forming a quantity approximately proportional to the derivative of the flux from the first component of the modified EMF vector or the magnitude of the flux vector and magnitude forming means for determining the magnitude of the correction vector.

22. Apparatus according to claim 20 or 21 and further including means forming a control vector acted upon by the frequency of the flux vector or the rotating reference system and preferably also by a quantity characterizing the load condition of the machine, which fixes, in functional dependence on its input quantities, the components of a control vector in the rotating reference system, and a multiplier stage which furnishes the correction vector by multiplication of the control vector by a magnitude determining quantity.

23. Apparatus according to claim 18, wherein means for forming an angle signal includes an angle servo control which is connected ahead of the second integrator and has as an input the difference between the angle of rotation and a control rotation angle for forming the feedback signal; a magnitude servo control acted upon by the first integrator signal and a control amount forming a further feed back signal and said computing stage including an adding stage for summing the further feedback signal and the first orthogonal EMF
component.

24. Apparatus according to claim 23, wherein said integration stage contains a divider for forming the quotient from the second orthogonal EMF component and the first integration signal, and wherein the output signal of the divider is fed to the input of the second integrator and the second integration signal to an output device for the angle or the Cartesian components Or the flux vector, and further including double-throw switching means for coupling the input of the second integrator selectably or alternatively between the output of the divider and the angle servo control.

25. Apparatus according to claim 18, wherein said EMF detector contains means for forming a smoothed EMF vector which is fed to the computing stage for transformation into the rotating reference system; the first integrator signal and a signal taken off at the computing stage output for the first orthogonal EMF component is proportional to the smoothing time constant form a first Cartesian component of the flux vector in the rotating reference system, and wherein a signal which is taken off at the computing stage output for the second orthogonal EMF component and is proportional to the smoothing time constant provides the second Cartesian component of the flux vector in the rotating reference system.

26. Apparatus according to claim 18 or claim 25, and further including means for transforming the flux vector into the stator oriented coordinate system.

27. Apparatus according to claim 18, and further including a device connected ahead of the voltage input of the EMF detector for suppressing d-c voltages.

28. Apparatus according to claim 18, and further including means for feeding the angle of rotation of the rotating coordinate system to means for the coordinate transformation of actual and/or reference values, and wherein the transformed actual and/or reference vectors of the control device are fed to a converter supplying the stators current of the rotating field machine.

29. Apparatus according to claim 18, in combination with a converter fed rotating field machine, wherein the position of the flux is determined and the converter is controlled in dependence on the determined position of the flux, such that the component parallel to the flux and the component perpendicular thereto of the stator current can be influenced independently.