EP0870358A1

EP0870358A1 - Compensated field-orientated control (kfo)

Info

Publication number: EP0870358A1
Application number: EP96946121A
Authority: EP
Inventors: Rupprecht Gabriel; Peter Magyar
Original assignee: D-Tech Antriebstechnik und Mikroelektronik GmbH
Current assignee: D-Tech Antriebstechnik und Mikroelektronik GmbH
Priority date: 1995-12-29
Filing date: 1996-12-30
Publication date: 1998-10-14
Also published as: DE19681187D2; WO1997024796A1

Abstract

The invention relates to a process and a circuit for field-orientated control of an induction machine (DSM, DAM). During the process, the flux angle of the machine or generator flux and/or the angular velocity thereof (the flux angle comes from the angular velocity as a result of integration) is determined in that an induction machine (DAM, DSM) transverse flux (frq=f(uid_err)) calculated while the machine (DAM, DSM) is in operation is adjusted (20, 30) to zero.

Description

Compensated field-oriented regulation (KFO)

The invention relates to a control (method and circuit / device) in field coordinates for three-phase machines, in particular for three-phase asynchronous motors (DAM) or three-phase synchronous motors (DSM).

Regulated electrical induction machines (i.e. synchronous and asynchronous machines) are used in many drive tasks. So far, these machines have been fitted with frequency inverters with characteristic control, provided the dynamic requirements are low, or with field-oriented controls and attached to the motor

Encoders or resolvers operated when the demands on the dynamics are high. The invention relates to a method for determining the transformation angle p (rho) of a field-oriented induction machine from quantities measurable at the connections of the machine. As a result, the encoder can also be saved in highly dynamic systems and the dynamics can be improved compared to previously known methods for sensorless control (without encoder or resolver). Compared to known stator flow-oriented methods, operation up to and including standstill of the machine is possible without singularity problems, as are known hitherto. The process is of great economic importance because it offers the possibility of expanding the field orientation to drives without sensors. In many drives, encoders today represent a very high proportion of the costs of the overall system. Nevertheless, the proposals according to the invention can also be used in the case of previously known field orientation of DSM and DAM.

The field-oriented control (see, for example, Blaschke [1], Leonhard [2]; more detailed information is at the end of the description) of the induction machine has proven to be a superior control method for induction machines in the prior art. By transforming the current and voltage stator variables into a coordinate system rotating with the rotor flux pointer (short: f _r ), a decoupled block diagram structure of the induction machine results, which corresponds to that of a compensated DC machine. The known control methods for such machines with, for example, cascade control for current, speed and flux can thus be transferred to the induction machine. The majority of all highly dynamically operated induction machines are operated according to the field orientation method - often also referred to as "vector control".

The basic problem of field orientation is the determination of the spatial pointer of the river in the machine. In the early days of field orientation, attempts were made with little economic and technical success to direct the flow in the machine, e.g. using semiconductor sensors. With the advent of high-performance microcontrollers and signal processors, it became possible to solve the coordinate transformation task and the highly nonlinear flow models based on the rotor equation digitally in real time, which is difficult to solve analogously (see e.g. Gabriel, Leonhard, Nordby, Schumacher in [3], [ 4], [5]). With the triumph of digital signal processing, the economic success of field orientation also began as a principally superior process.

Early attempts were made to use the stator differential equation to determine the flux pointer. The open integration that is used here, however, prevented use in and near standstill, because parameter errors in the determination of stator resistance R _s and leakage inductance L _s, and in particular the offsets in the components of current and voltage occurring during the measurement, lead to one lead to unpredictable drift of the calculated components of the flow in the machine. It has therefore been assumed that the flow can no longer be observed at this singular point near the standstill.

The method called "natural field orientation" (NFO) (cf. Jönsson [6], [7]) can work without a sensor, but does not generate a field orientation and thus completely decouples the structure, which neglects the most important influencing factor for determining the angle p (rho) of the river. The "natural field orientation" thus only represents the transformation of a characteristic curve method. A subsequent field orientation is not achieved. The problem of the invention is to propose a control method that can safely control three-phase machines even in the standstill range without rotary encoders or angle encoders (“sensors”). According to the invention, the field orientation is achieved by regulating the cross flow of the machine to zero is the compensated field orientation (KFO) (claim 1, claim 13.) This method for determining the flow angle (rho or p) solves the aforementioned problem without a principle-related singularity and is suitable for field-oriented operation from standstill to field weakening without support from an encoder.

Experience, especially with sensorless field-oriented control of three-phase motors, has shown that a control method according to the state of the art (see page 14 in [11], [12] or [13]) leads to insufficient results being achieved in certain work areas were. The reason for this is that accumulated errors, caused by parameter errors, measurement errors of the signals, calculation errors, etc., result in a small but not negligible misorientation. A resulting cross flow in the machine deteriorates the dynamic properties and leads to instability, which has been the main hindrance in the use of sensorless vector methods, since the neglect of this cross component is a lack of information about the system state. The invention is based on the idea that this _{transverse component} f _{rq is} actually regulated to zero - instead of being neglected (claim 1, claim 13).

In a rotating coordinate system with the rotational speed ω _mr , f 1 can be determined from the stator equation 1 and the rotor equation 2.

The following applies to the stator equation of the induction machine:

,. ". di. ... ., German

1. u _s = ι _s R _s + L _s - ± + jι _s L _s ω _mr + jf _r ω _mr + - £ υ _s - stator voltage vector / _s - stator current vector

R. - stator resistance - leakage inductance

Rotation speed of the rotor flux pointer fr of the motor model dp dt f _r - rotor flux pointer j + VM, imaginary unit

The following applies to the rotor equation of the induction machine:

Hf

2- 0 = - ^ - + i _r R _R + jf _r ω _r

R _R - rotor resistance ι _r - rotor current vector ω _r - rotor rotational speed compared to the flux vector ω - ω _mr - ω _r rotational speed of the shaft

The synchronous machine DSM can be regarded as a special form of the general three-phase machine with R _R = 0 and existing flux f _r . The method according to the invention can therefore also be applied to synchronous and permanently excited induction machines.

Stator equation 1 consists of two component equations:

Hi Hf 3. u _d = i _d R _s + L _s - ^ - - i _q L _s ω _mr -f _π , ω _mr + - ^ -

_Λ di _a df _m

4. u _q = ι _q R _s + L _s - ^ - + i _d L _s ω _mr + f _rd ω _mr + - ^ - where the indices d and q denote the direct and orthogonal axis of the coordination system rotating with f _r and u _d and u _q the d and q components of the voltage vector u _s , i _d and i _q the d and q components of the current vector i _s , f ^ and f _rq the d and q components of the Are rotor flux pointers f _r , where f _{r (J is} the main flow and f "is the cross flow. The voltage components u _d and u _{q of} the machine are composed of the voltage drops and the voltages caused by changes in current or flow, and the voltages induced by the two flow components f _rd and f _rq . The two induced voltages are defined by the equations

U = fr, * CO, ^d f "> iq J ra mr H i.

The expressions for the induced voltages calculated from the Sator equations result from equations 3 to 6.

di 8. u _ιq = u _q - i _q R _s - L _s - ^ - - i _d L, ω _mr

Equations 7 and 8 of the induced voltages show that, as is known, these can be calculated from the voltages at the terminals of the machine and the deduction of the stray inductances caused by the resistances and inductive voltage drops and the voltage drops caused by the change in flow. The equations in the stator or in rotating dq systems can now be calculated in real time using fast microprocessors and signal processors. u _ιq can be calculated in a coordinate system with f _r as well as in the stator system with subsequent transformation into a dq coordinate system (Jönsson, Leonhard [8]). In the field-oriented state, f _r q must be zero by definition.

9. / ^ = 0

So you get 10. u, _q = / „, • ω _mr 11. df, approx

^Λ , d dt

The two equations 10 and 11 are used to determine the field orientation. The equation 10 gives the estimated value for ω _m r as follows:

The signal Ujd _err for the error voltage of the induced longitudinal voltage Ujrj can be calculated from equation 11, which represents a measure of the quality of orientation:

13. u _id _err = ^u * - ^ jf

This signal represents a measure of the compensation quality. In the case of the compensated field orientation, the signal ujrj _err = 0, otherwise it results from

Equation 5 an orientation error.

14. Uj ^ err = - ^f _rq 'ω _mr

A transverse flux f _r q characteristic of the orientation error can be calculated from this, in accordance with equation 15.

15. f _rq = - ^^ mr

Using equation 12, this characteristic cross flow is described as equation 16.

16. f _rq = - V ^ / ", The meaning of equation 16 was recognized for the correction of the slip frequency ω _r (see, for example, Rapp [9], [10]]). This is an indirect method with which the errors in the slip frequency and in the identification of R _R can be reduced. The method is not suitable for sensorless controls because the transformation with p does not take place with a corrected value.

The rotational speed ω _mr is determined directly by the method according to the invention. This is done by regulating the cross flow f _rq to zero. Here, f _{rq is determined} from u _ld , u, _q and f _{rd in} accordance with equation 16. This is achieved, for example, by _{calculating a quantity} proportional to _rq from u _ld , u, _q and f _rd .

The invention can be explained specifically using equation 16. If the cross flow f _{rq is} to be regulated to zero - only then is the field orientation really given - then f _rq can be _calculated from u, _d and u _ιq . It is known from f _rd that it should be controlled positively and to a value specified by the field transmitter, which is necessary for the magnetization. An incorrectly calculated f _rd is therefore not a fundamental obstacle to regulating f _rq to zero. The invention makes it possible to enable field-oriented operation even with relatively poorly adapted flow models.

The compensated control is not an "observer" and has no observer feedback. KFO works continuously without switching or changing control loop structures. KFO works with a d / q component structure that is closest to the flux / torque structure of a three-phase machine. In orthodontic errors, orthodontics are corrected at the root, especially before the

Formation of the Rho transformation angle, which is the linchpin of the field orientation. Its accuracy - in the long term and at critical operating points, such as standstill and strong field weakening - guarantees an exact field orientation (Rho orientation), so that sensorless operation is also possible with induction machines. The implementation of regulating f _rq in a stator model of the field orientation to zero or keeping it around zero (in the case of briefly permitted deflections but continuous control behavior aimed at regulating the deflection) in order to achieve the compensated field orientation is shown schematically in Figures 1 and 5, as exemplary embodiments.

A function comparison with NFO (cf. [7], [8] according to page 14) is shown in Figures 6a, 6b, in each of which an asynchronous servo motor starts up with a 1.5kW nominal power at a nominal frequency of 50Hz and a nominal torque of 5Nm be shown, once for the KFO and once for the NFO.

Examples explain the invention with reference to pictures (= figures).

Figure 1 is a schematic model of cross flow control. Since the controller 20 is _intended to regulate the cross-flow f _rq to zero, a greatly simplified sliding mode controller is also suitable, which only takes into account the signs of u _{ld err} and u _ιq . This eliminates the division's solution, which is complicated for practical implementation in arithmetic _units and also _unique at u _ιq => 0. The controller should have a high dynamic, since p results from the integration 40 of ω _mr and a following error can lead to large angular deviations and thus even to the machine tipping. Lag errors can be avoided by means of pilot controls or fast controllers.

Figure 1a illustrates the calculation of the error u _{ld err} , from which the cross-flow results, which is regulated to zero (around zero) in Figures 2 and 3.

Figure 2 shows a circuit for compensation of the cross flow. The controller 30 is preferably a proportional controller, which in the simplest case can also have the gain Kp = 1. The integrating controlled system 40 ensures that there are no permanent control deviations even with a proportional controller.

Figure 3 is another variant of the determination of the flux vector angle p (rho). The

Another advantage of this solution, in addition to the almost equally simple implementation according to Figure 2, is that errors in p have a smaller intervention size and are therefore gentler

be settled. The signal - - here represents a pilot control ω _{NF0 of} the NFO

J rd

The controller 20 only has to intervene (in addition) via ω _sκ if the model of the machine does not match reality. Especially with a superimposed speed control with a slip model

/ R ω = ω _mr - co _r and ω _r = - - - it is helpful not to interfere too much with the f ^ -

J rd

Controller 20 to have. Figure 4 corresponds with its compensating circuit to the circuit of Figure 3, only the comparator 10 is designed as a comparator-like function block 11 with a large but finite slope around the zero point.

Figure 5 is an overview of the relevant components of a field orientation with DAM induction machine and PWM inverter. The circuit from Figure 4 is marked with XR in Figure 5.

Figures 6a, 6b are comparison plots of orthodontics with NFO, with a simulated startup.

In an f _rq model 9, an at least the sign of f _r - proportional model of f _rq = f (u _ld , u _ιqι ) is formed in Figure 1. A suitable subsequent controller 20 or 30 can then regulate f _rq to zero. The setpoint size of the controller is zero, so f _rq ^* - (the at least correct sign f _rq ) must be inverted (added). ω _mr is therefore not from u _ιq / f _rd or worse from that assumed to be constant

To calculate main field inductance. Each error is accumulated by the subsequent integration of ω _mr to determine p and therefore leads to a non-field-oriented system. _Controlling u _ld to zero would also be an insufficient description, since f _rq is essentially determined by the ratio u _ld to u _ιq .

It is possible to integrate the integration for ω _mr in the controller 20 or 30.

f ^ is the flux in a field-oriented system and is also referred to as rotor flux f _r if f ^ is zero. In a system that is not adjusted, the cross flow is not zero and the field orientation is not fulfilled either. If the cross flow, which is determined from u _ld and u _(q , is regulated to zero, the result is a completely compensated - the cross flow is compensated by a controller - field orientation.

The controllers 20, 30 in Figures 2 and 3 can also be used with proportional controllers

Gain of Kp = 1 or greater. The particularly simple case Kp = 1 has brought good results in most cases. Overall, the structure of the controllers, especially in the structure of Figure 3, not critical. Both P, Pl and PID controllers are suitable for the f _rq controller 20 or 30.

An effect that has not yet been shown is the behavior in the case of parameter and measurement errors in the determination of the induced voltages. In many cases, the

Calculation of u, _d and u _{ιq are} simplified. In the case of low dynamic requirements, the voltage drops at the scatter and the voltage drops caused by the change in flow can be dispensed with. In practical application it is not _possible to determine completely free of errors u _ld and u _ιq . Such errors then lead to a rough running of the machine near the

Standstill. In order to minimize the effects of the errors, it can make sense to replace the two-point controller (Signum function) in Figures 2 and 3 with at least partially approximately linear controllers with finite gain and limitation, as illustrated in Figure 4. Small errors in f ^ will therefore have no major interference in ω _mr near standstill. However, in the case of zero speed, for example, this means that ω _{mr is} no longer determined correctly in the case of a non-integrating controller. In the closed control loop, the speed of the machine is no longer observed correctly and thus a slight misorientation can be accepted.

It is of course the case that the avoidance of errors in the determination of u _ld and u, _{q is} always advantageous for the speed and torque linearity of the system. The identification of R _s and L _s and an accurate measurement of the currents and voltages are therefore advantageous. However, the invention makes it possible to compensate for remaining errors. In addition to linearly limited controllers, non-linear controllers can also be used. The induced voltages can be calculated from the measured stator voltages and the corresponding equations 1 or 5 and 6 voltage drops to be calculated in a stator or rotor or else in a mixed manner.

Usually the stator voltage for calculating the induced voltage (by measurement) is known. You can't avoid a current measurement Voltage measurement can be bypassed, however, since they have no internal dynamics other than the operating times of the circuit breakers, which, however, can mostly be neglected. With fast switching converters, the voltages at the converter output can be influenced not only by the DC link voltage but also by the dead times and voltage drops of the switches. Dead time compensation is therefore necessary in many cases if, instead of the voltage measurement, the degree of modulation multiplied by the intermediate circuit voltage (which should be measured or is known as a constant) is to be used as a replacement for the voltage measurement.

Figure 5 shows an example of a complete field-oriented control of the induction machine (DAM) with an ω _mr or p determination according to Figure 3. It shows the ω _mr determination according to Figure 3 with XR as well as the known control loops and models for control in a field-oriented system, f ^ is obtained from an i _sd flow model FM, as used in controls with rotor position sensors. The control includes the current controllers 50d, 50q for i _sd and i _sq known from the field orientation and superimposes the field and speed sensor 60. The actual value of the speed controller is determined by a speed monitor DB. Here ω _r or the slip frequency is determined from a rotor model via ω _r = i _sq ^• R _R / f _rd .

The result is an operation equal to that of the DC machine, even around standstill. In a synchronous machine (DSM), ω _r = 0. A good knowledge of u _jd and u _{iq is} helpful for an accurate identification of ω _mr and p. Angular errors are caused exclusively by errors in the determination of u _jd and u _jq . However, the errors can be neglected by measuring voltage, current and parameter adjustment of R _s , L _s . This method is therefore suitable for sensorless field orientation even for position-controlled drives (servo drives).

Figure 6a is a ramp-up of the asynchronous servo motor mentioned at the outset, showing the main machine sizes. Figure 6b is a comparative example, 13

where the same machine sizes are shown graphically, only with a different field-oriented control than in Figure 6a. The latter figure shows the mechanical angular velocity ω _{s of} the rotor and the stator voltage u _s in the upper or lower field when the servo drive is operated with the compensated field-oriented control (stator model for field orientation) according to the circuit in Figure 3 or Figure 4 in the system circuit diagram 5. The target speed ω _thus “increases from zero to three times the nominal speed at approximately 50 msec, with approximately 20% remaining flux in the field weakening range. Even at the beginning of the second field weakening range (reaching the current limit of i _q (t) with decreasing torque m _mo1 (t)), the KFO continues to show stability, while the NFO in Fig. 6b begins with considerable vibrations in the stator voltage u _s To clearly show orientation errors of the flow angle p (rho). Even in the third field weakening area, where the servo motor works at the slip limit, the orthodontic control system in Figure 6a remains stable, while the NFO control system shows a particularly high tendency to oscillate.

The operating parameters of the rotating field motor are identified in the same way in both figures 6a and 6b, and the time axis t (time) also has the same scale. So a direct judgmental comparison is possible. The signal curves in the pictures mentioned were created with a sampling frequency of the controller of 20 kHz. A is shown

System with measurement of the machine signals necessary for the control. It is easy to imagine that the tendency of the NFO system to vibrate is further increased at the critical points in the second and third field weakening range when working with more sensitive sensorless control of the servo motor. The parameters in Figure 6 are:

The mechanical angular velocity is ω _s (t) The angular velocity of the rotor flux f _r is ω ^ t) The nominal value of the angular velocity is ω _sol | (t) The stator voltage is u _s (t) The cross current is i _q (t), the longitudinal current is i _d (t)

The engine torque is m _mot (t). State of the art:

[I] Blaschke, F .: The method of field orientation for controlling the asynchronous machine. Siemens Research and Development Reports 1 (1972) 1, pp. 184-193. [2] Leonhard, W .: Microcomputer in electrical drive technology. Interkama Congress 1980, p. 637.

[3] Gabriel, R., Leonhard, W., Norby, C: regulation of the converter-fed

Three-phase asynchronous machine with a microcomputer. Control engineering 27 (1979) 1,

Pp. 379-386.

[4] Gabriel, R .: Microcomputer-controlled asynchronous machine, a drive for high dynamic requirements. Control Technology 32 (1984) 1, pp. 18-26.

[5] Schumacher, W., Leonhard, W .: transistor-fed AC-Servo Drive with

Microprocessor control. Proc. of IPEC, Tokyo, 1983, pp. 1455-1476.

[6] Jönsson, R .: in WO 91/12650, US Patent 5,294,876 (NFO).

[7] Jönsson, R .: Direct, Indirect and Natural Field Orientation (NFO) for Control of the AC Induction Motor. Manuscript, 1995.

[8] Jönsson, R., Leonhard, W .: Control of an Induction Motor without a Mechanical

Sensor, based on the principle of "Natural Field Orientation" (NFO). Proc. Of IPEC,

Yokohama, 1995, pp. 298-303.

[9] Rapp, H .: Examination of Transient Phenomena in Induction Machines, caused by an Incorrectly Adjusted Rotor-Time Constant in a Field-Oriented Control System. ETEP

Vol. 3, No. 6, Nov / Dec 1993, pp. 397-405.

[10] Rapp, H .: in EP 92 10 6811 (EP 570 602) and P 42 09 305.

[I I] Blaschke, F .: in EP 127 158, in particular FIGS. 1 to 4 and pages 7, 8 there, but without any real regulation of the transverse flow f ^ neglected there. [12] Blaschke, F .: in US 4, 626,761, in particular FIG. 1 there and column 4, where a

An observer is described who simulates the flow angle in polar coordinates. [13] Jötten / Mäder: in DE 33 32 567, there Figure 3, function element 317 and Figure 1, controller 12, but both of which work in the rotor flux system, the Signum function 15 working there in the feedback of the controller 12. There ω. Is not the angle of rotation of the flux vector, but that of the rotor EMF.

* * *

Claims

Expectations:

1. Method for field-oriented control of a three-phase machine (DSM, DAM), at

which is the flow angle (p) of the machine or generator flow of the machine

and / or its angular velocity (ω _mr ) - where the flow angle (p) through

Integration (40) _results from the angular velocity (ω _mr ) - determined thereby

is that a transverse flux (f _rq = f (u _{ld err} )) of the induction machine calculated during operation

(DAM, DSM) is regulated to zero (30.20).

2. The method according to any one of the claims mentioned, in which an at least partially

approximately linear function element (10, 11) with or without limitation for

Control of the controller (20,30) is used, in particular a Signum function to

to determine a control difference corresponding to the sign for the regulation.

3. The method according to claim 1 or 2, wherein the angular velocity (cO by

u addition of -j- and the output signal of the controller (20) is determined, the

J rd

Input signal essentially the product signal (u _{ld err} x Signum u ^)

Sign of the transverse stress and one corresponding to the cross flow

Error signal, especially in the timing of a scan control of the machine (DAM,

DSM).

4. The method according to any one of the preceding claims, wherein the angular velocity

(ω _mr ) from the output signal (ω _sk ) of the controller (20), the input of which is a second

Product input signal (signum u _{ld err} xu _ιq ) is supplied, which is in the

essentially from the sign of the error signal and the transverse _voltage (u _rq )

calculated.

5. The method according to any one of the claims mentioned, in which the

Angular velocity (ω _mr ) the output signal (co _sk ) of the controller (20) with the

^u , d input signal - —f ", where the cross-flow (f _rd ) from a flow model

(FM) with the longitudinal current (i _sd ) as an input variable, and a

Pilot signal (u ^ / fj to the controller output signal (ω _sk ) is added to the

To form angular velocity (ω _mr ).

6. The method according to any one of the claims mentioned, in which the controller (20,30)

P, Pl or PID controller and / or the signals are analog or digital signals.

7. The method according to any one of the claims mentioned, in which the

Angular velocity (ω _mr ) is the output signal of a sliding mode controller,

at which with positive cross flow (f ^) the output signal (ω _mr ) increases and at

negative cross-flow (f _rq ) is reduced.

8. The method according to any one of the claims mentioned, in which the

Angular velocity (ω _mr ) by adding a pilot signal (ω _mr .)

proportional to the quotient signal (u, _q / f _rd ) from transverse voltage and

Longitudinal component of the rotor flux pointer is obtained and a

Sliding mode controller is used.

9. The method according to any one of the claims, wherein the speed (ω) of the

Machine (DAM, DSM) as the output of a speed monitor (DB)

^ω _r - ^ω _mr ^{* is} obtained;

'«7 Rn where ω _r from - ^ - - (L as cross current, R _R as rotor resistance, f as output of the

• approx

Flow model with i _sd as input variable) is obtained;

where ω _mr ^* from ω _mr and a multiplication by - the amount of

Output signal of an at least approximately linear function element (10, 11)

corresponding - multiplication factor is obtained;

a normalization takes place so that the multiplication factor at the limits

corresponds to exactly one.

10. The method according to any one of the claims mentioned, wherein

the induced voltages (u _id .u _jq ) of a model (SM) by measuring the

Stator voltages and currents and calculated in the stator or rotor system;

the calculated cross flow (f ^) is the cross component of the flow vector (f _r ).

11. The method according to any one of the claims, wherein the stator voltage from

the target values for a modulation (PWM) and a measured or estimated

DC link voltage of an inverter by multiplying the

Degree of modulation can be obtained with the DC link voltage.

12. The method according to any one of the claims mentioned, in which from an induced

Longitudinal error voltage (u _{ld err} ) and an induced transverse voltage (u _iq ) with the

The sign of the cross flow (f _rq ) corresponding signal is obtained,

which directly or inverts the input signal of the cross flow - in operation -

around the controller (20.30).

13. Device for performing the method according to one of the mentioned

Claims where

(a) a controller (20, 30) is provided which is the one in the model (XR)

cross-flow (f _rq ) to zero or nearly

Constantly regulates zero;

(b) the controller (20, 30) via a the sign of an error signal (u _{id err} )

evaluating comparator-like signal stage (10) is fed, the

Error signal from longitudinal voltage, transverse _voltage (u _id , u _jq ) and sign

of the cross flow (f _rq ) is formed.

14. The apparatus of claim 13, wherein a calculation unit is provided, the

the sign of the error signal (u _{ld err} ) acc. Claim 12 (without his

Return).

15. The apparatus of claim 13 or 14, wherein the signal stage (10,11) acc.

Claim 2 (without its back reference) is formed.

16. The method or device according to any one of the preceding claims, in which

the output signal of the controller (20,30) is directly proportional to the

Angular velocity (ω _mr ) of the flow vector (f _r ) or only directly proportional to

a supplementary _signal (ω _sκ ), which becomes a suitable pilot signal (ωJl),

especially the angular velocity signal (ω _NFO ) of the flow vector of the NFO

Field orientation is added.