DE10038570A1 - Stator current regulation in electrical machines uses sampled values with delay in PI controller - Google Patents

Abstract

The control of the current in the stator of an electrical machine is based upon sampling of the actual stator current and comparison with the reference value . This is used in a PI controller. The system employs a delay between successive sample values to provide an average value that is integrated.

Description

The invention relates to a method for regulating the stator current electrical machine, wherein an actual value of the stator current is detected and with a Current setpoint is compared, and one from the setpoint / actual value comparison resulting control deviation using a control device to a Control output value for a control voltage of the stator or at the Machine is implemented. With discrete time control of the stator current with discrete sampling times for individual actual values of the stator current Control output value by one of two consecutive sampling times Defined sampling time interval delayed as control voltage on the machine spent. The invention further relates to an I control channel for a PI, PID or other controller for use in a current control loop Stator winding of an electrical machine, the I control channel in particular is suitable or designed to carry out said method. Further The invention relates to a corresponding PI controller and a current control loop the stator winding of an electrical machine, the one with current setpoint and Current / actual value reference and actual value comparator and one of these Comparator downstream PI controller.

Regulations of the stator current of an electrical machine of the type mentioned are known. Usually a PI controller is found because of its relatively good Properties and its ease of use in the generic Drive technology (DC and three-phase machines) application. Especially in There is a considerable need for fast or highly dynamic and at the same time stable and precise working Control loops in which the increase in the stator current can be maximum. If necessary, overshoot should also be contained or avoided.

To achieve this object, that specified in claim 1, Discrete-time control method for the stator current of an electrical machine proposed that a current I-regulation Control deviation and an older one by at least one sampling interval  previous control deviation can be averaged, and a resulting mean value over time for the control output value is integrated. Numerical are of course used for discrete-time or digital control Integration process used. According to the invention with the help of Dead time element at least also a past control deviation in the current one Sampling interval is taken into account, the accuracy of the I control channel and in the case of a combined control, for example PI control, the Increase the accuracy of the stationary component. Any Pulse width modulation, which follows the control as an actuator, can stabilize with the invention.

A practical, computationally simple realization of the averaging consists of adding the current and older control deviation together and then the addition result before further (numerical) integration cut in half. The current control deviation can, before it is input variable for the Averaging forms over proportional links with one or more Control parameters multiplied, linked or otherwise processed or also be directly fed to the averaging.

The stability of the scheme promotes if, after a different training of the the method according to the invention the control deviation before averaging it Amount or size reduction is subjected, which only then is performed when the initial value of the control is a predetermined one Limit exceeded.

To solve the invention task mentioned at the outset is to regulate the Stator current of an electrical machine using a combined P and I control proposed according to the invention for the setpoint / actual value comparison predictive current actual value to be used by adding the stator detected current actual value with a change in the determined by current prediction Current actual value is formed. The electricity prediction is known per se. Here will but proposed a new combination with a PI controller. This  Alternative invention is not based on the use of time-discrete, digital Control systems with sensing elements limited, but for such controls particularly suitable.

So is a special training of this invention Process alternative on a discrete-time control, in which in discrete Sampling times, the actual value of the stator current is detected, and that over the Control method immediately calculated control output value by one Delayed sampling time interval is converted into the control voltage for the stator. The special training is characterized by the fact that within the Current prediction a predicted stator current change with the last one sampled actual stator current value for the subordinate setpoint / actual value comparison is added. In particular, this will be designed in a time-discrete manner Regulation for the addition of the actual stator current value used in each current sampling time was recorded, whereas for the current prediction Control output value is used, which is in a sampling time interval before current sampling time was calculated. The used according to the invention Current prediction is inherent in the idea of one for a subsequent sampling time interval Change in the current actual value to be forecast and the last sampled Current actual value to add, and then the resulting addition result with the Compare current setpoint.

As part of the general inventive idea, the solution to the initially mentioned invention task further (not on time-discrete or digital I-control channel for a PI, PID or other Regulator proposed for use in stator current controls that stand out a device for forming an average from a current and at least an older control deviation, whereby the averager before Integrator of the I control channel is arranged. With this I control channel the above-mentioned, discrete-time control procedure is appropriate realize. In order for digital, time-discrete control of the system-related Delay of a pulse width modulator as an actuator by a sampling time interval  or to meet a sampling period is after a special training The invention proposed that the dead time of the dead time element of the I control channel is dimensioned according to the sampling interval of an external sampling element, which for Detection of individual, discrete current actual values is used. The entrance of the I- Control channel is designed for connection to a setpoint / actual value comparison element, which is fed by the sensing element with the discrete actual current values.

As part of the general inventive idea to solve the problem mentioned invention task also a PI controller for use in a Current control loop of the stator winding proposed which controller next to that P control channel has an I control channel, which after and above Invention principle specified claims is formed.

The I control channel of the PI controller according to the invention is of a special design with a reduction device for input values or control deviations which limiting device the dead time element and the adder precedes and depends on the size or amount of an initial value of the controller or specifically the I control channel can be activated. The activation can for example by means of a threshold value decider known per se, its threshold or trigger level to a maximum limit of Controller output value is dimensioned.

To solve the invention task mentioned at the outset and / or Implementation of the above-mentioned regulatory procedure with electricity prediction According to an alternative of the invention, a current control loop with comparator and downstream PI controller proposed, in which a device for Current prediction is provided. This current prediction generates from Input values such as the controller output value and the current actual value and if necessary, an actual EMC pre-control value as a change in current value resulting output variable. The current prediction device is on the output side indirectly connected to the comparator via a summing element to which the Current actual value is supplied. This opens up the possibility according to the invention  carry out a target / actual value comparison, in which the actual value is a predictive one Component, which is also a predictive component in the management of the Control loop results.

If the current control loop is implemented as a discrete-time scanning system, where the control output variables are always delayed by a sampling interval in the stator control voltage can be implemented, it is appropriate that the Prediction device comprises a dead time element with a dead time constant that the Corresponds to the sampling time interval. The dead time element is then the initial value of PI controller or, in the case of EMF feedforward control, a subtraction result between the controller output value as a minuend and the EMF pilot control value fed as a subtrahend. The actual stator current value of each current sampling time interval with those delayed according to the dead time element Controller output values for a predictive current actual value prediction processed further.

Further details, features, advantages and effects based on the Invention result from the following description more preferred Embodiments of the invention and from the drawings. These show in:

Fig. 1 shows the structure diagram of a current control loop for the stator winding of an electric drive, as is known in the art,

Fig. 2 shows a current-time diagram for the invention to be aimed current profile,

Fig. 3 shows a current-time diagram to be aimed for at a high speed or with a large inductance of the electrical machine stator current profile,

Fig. 4 is a structural diagram of a PI current regulator according to the invention,

Fig. 5 shows a power-time diagram of the stator current waveform with that in Fig. 4 shown current controller,

Fig. 6 is a power-time diagram for current waveforms with the current regulator according to the invention, additionally provided with a current prediction,

Fig. 7 shows a structural diagram of the current controller of the invention with current prediction,

Fig. 8 - Fig. 11 step responses of the stator in accordance with the invention and of prior art control for different speeds.

A known standard current control loop in an electric drive can be generalized as shown in FIG. 1. The motor winding can be described approximately electrically as a first order PT1 delay element if an induced voltage ("motor EMF") is compensated for by a pilot control known per se. In particular, according to Figure 1, the stator winding forming Fig. 1 the control path to said transfer member PT1 and an upstream proportional term 1 / r _S. r _S stands for the resistance of the stator winding. The controlled system 1 also contains a summing element 2 connected upstream of the proportional element 1 / r _S , which expresses the influence of the motor EMF emk_Motor, which is proportional to the speed. The stator winding or controlled system 1 receives its input or control voltage usq from a pulse width modulator 3 , which is symbolized in FIG. 1 as a dead time element 3 with a dead time constant Tab. The dead time constant has the same amount of time Tab as the sampling time interval Tab of a sampling element 4 illustrated as a switch in FIG. 1. The latter detects a discrete actual stator current value isq at equidistant sampling times, which is fed from the output of the sampling element 4 to a nominal / actual current comparison element 5 with a negative sign. The current control deviation Δi, which is fed to the PI controller 6, results at the output of the comparator 5 . In a known manner, this consists of two control channels connected in parallel, a P control channel with the control parameter kp and an I control channel with the parameter kp / Tn. An integrating element 8 is connected downstream of the proportional element 7 containing the latter I control parameter in the I control channel. The resulting two control channel output values u_P_Teil, u_I_Teil (for P and I components) are fed to a controller output summing element 9 . Another input of this output summing element 9 is connected to a pilot control 10 which, in addition to the speed / speed speed, works as the main input variable with a standard value NORM_EMK for a "normal EMF" (standardization factor for electromotive force) and with the magnetic flux psi as further input variables. The latter are linked to the speed speed via respective multipliers 11 , 12 . As already mentioned, the actual current in the stator winding is sampled at equidistant sampling times in each case after the sampling interval Tab by means of the sampling element 4 in the digital control system shown. From this, the control output value usq_ref is calculated via the PI controller 6 , which is subjected to a limitation 13 at the output of the controller output summing element 9 and is supplied to the pulse width modulator 3 as an input variable. This converts the control output value usq_ref into a control voltage usq, which is generated anew with each sampling time or sampling time interval Tab. A change in the control voltage usq is possible due to the digital scanning system with the scanning element 4 only in the next or following scanning time interval Tab and can only be output to the motor with a delay of one scanning time interval Tab. This means that the pulse width modulator 3 forms an additional delay element or dead time element in the control loop, which must be taken into account when designing the current regulator.

In order to achieve a highly dynamic and stable servo drive, the following requirements are placed on the current controller in the control loop shown:

• - Maximum increase in current
• - No overshoot.

The optimum amount is known in drive technology for determining the parameters of a current controller. The delay element or dead time element 3 is assumed to be a PT1 element with a time constant of 1.5 times the sampling interval Tab. Taking into account the pilot control 10 in relation to the induced voltage, the dimensioning of the PI controller 6 is carried out only according to the guidance behavior. How to get the transfer function of the open control loop:

where:
r _S resistance of the stator winding
T _{from the} control interval of the circuit, or sampling time,
T _S = I _S / r _S the electrical time constant of the stator for synchronous machine,
T _S = I _σ / r _S the electrical time constant of the stator for asynchronous machines
I _S is the stator inductance of the synchronous machine and I _{σ is} the total leakage inductance of the asynchronous machine.

The gain and reset time are based on the optimum amount

With T _S = I _S / r _S you get the P and I factors as parameters for the PI controller:

The properties of the control loop with the optimum amount result as one Compromise between dynamics and overshoot. However, it is impossible with to obtain a control loop that is characterized by high Dynamics and responsiveness on the one hand and only through minimal Overshoots or lack thereof. The large Kp factor ensures a PI controller for dynamics / responsiveness. A fitting Reset time Tn is decisive for the control accuracy.

In order to achieve the desired properties for the current control loop, a controller must be able to correct the control deviation Δi = isq_ref-isq in a sampling time interval Tab when the current setpoint isq_ref changes. Fig. 2 shows the waveforms of the desired current command value and the actual current isq_ref isq.

Assuming that the induced voltage is initially equal to zero, the required voltage integral can be determined for the courses shown in FIG. 2:

As mentioned above, the P component ensures the maximum increase in current and the I component ensures the stable and precise steady state of the current, here the voltage drop across the resistor r _S. In the next sampling interval Tab, the current has reached the setpoint, so the P component responsible for dynamics is no longer necessary and the I component is equal to r _S Δi (k):

usq (k + 1) = u_P_Teil (k) + u_I_Teil (k) = 0 + r _S .Δi (k) (3b)

Comparing equations (2b) and (3a), it can be seen that the with the invention to be aimed for controller KP and KI factor three times as large as with a conventional PI controller - equation (2b). This will make the Interference in the current control loop quickly suppressed.

The induced voltage increases with increasing speed. As a result, the voltage available to build up the current becomes smaller. The voltage coming from the controller 6 , which would be required to correct the deviation in a sampling time interval, will then no longer be passed on to the motor, but will be cut off by the limitation 13 . This also happens when the inductance of the motor is large. In such cases, the current needs several sampling periods Tab in order to reach the setpoint or manipulated value. In order to achieve the curves shown in FIG. 3, the I component must also be reduced in the same ratio as the P component. The corresponding controller is shown in Fig. 4.

In Fig. 3 it can be seen that the current should increase in the first three periods at maximum speed, which is determined by a voltage reserve. At the end of the fourth period k + 4, the setpoint isq_ref is reached.

In order to achieve the ideal current profiles according to FIGS. 2 and 3, the PI controller shown in FIG. 4 is proposed. This differs from the conventional PI controller according to FIG. 1 by a combination of a dead time element 14 and an adder 15 connected downstream of it. The first input of the adder 15 is directly connected to the input of the dead time element 14 and in parallel to the output of a multiplier 16 , and the second input of the adder 15 is connected directly to the output of the dead time element 14 . The input of the dead time element 14 and one of the adder 15 are thus applied in parallel to the output of the multiplier 16 . The output of the adder 15 is fed to a fraction factor element 17 which acts as a proportional element. The combination of the dead time element 14 , the adder 15 and the fraction factor element 17 has the effect of an arithmetic averaging from the control deviations of the last and penultimate sampling interval, since the dead time constant of the dead time element 14 coincides with the sampling interval Tab and the fraction factor of the factor element 17 with " Are dimensioned 1/2 ", so the denominator refers to the two sample values used for averaging. The output of the factor element 17 is fed directly to the integrating element 8 . The multiplier 16 forms part of a limiting device, which additionally has a dividing element 18 and a switching device 19 . The first input of the multiplier 16 is connected directly to the output of the proportional element 7 , which links or multiplies control deviations Δi with the controller parameter kp / Tn. The output of the divider 18 is connected directly to the second input of the multiplier 16 . The first input of the dividing element 18 , which represents the dividends, is connected to the switching device 19 which, in the normal state, connects the dividend input with the output value u_P_Teil directly from the P control channel. This is also fed in parallel to the divisor input of the divider 18 . Represents a (not drawn) threshold value determines that the not yet limited output value usq from the entire PI regulator is greater than a threshold usq_lim the previously discussed limit 13, the switching element 19 a of the switching device 19 folded over from the output of the P-control channel and applied to the output of a further summing element 20 . Its first input is the constant threshold value usqlim of the limit 13 . Its second, inverting input is fed to the output of a further summing element 21 , which forms a sum of the immediate output value u_I_Teil of the I control channel or integrating element 8 and the output value USQ_EMK of the aforementioned pilot control 10 . The output of the summing element 21 is summed via the aforementioned summing element 20 with a negative sign with the limit value usq_lim and with a positive sign with a further summing element 22 with the output value u_P_Teil of the P control channel. The provisional output value usq_aus of the PI controller is then present at the output of the last-mentioned summing element 22, which is available as a final control output value usq_ref after passing through the limit 13 . At the output of the associated limiting device 16, 18, 19, 20, summing element 20, the difference is usqlim from the limit value and the sum (addition element 21) of the I-control-channel output value u_I_Teil and the pilot output value USQ_EMK one of the two inputs of the switching means 19 provided . If the provisional control output value usq_aus on the summing element 22 is greater than the limit value usq_lim, the position of the changeover switch 19 in the dividing element 18 (changed compared to FIG. 4) causes a break with the difference in the counter and the output value u_P_part of the P control channel in Denominator formed. This fraction should generally be less than one and is supplied to the multiplier 16 as a factor. If the preliminary output value usq is less than or equal to the limit value usq_lim, the output of the dividing element 18 is always at one, so that the control deviation Δi multiplied by the proportional element 7 with the control parameter kp / Tn from the multiplier 16 is practically unchanged at the dead time element 14 and at that one of the inputs of the adder 15 is transmitted.

By in Fig. 4 shown PI controller could be the ideal current waveforms in Fig. 2 and Fig. 3 are achieved when usq_ref the control value without the delay tab could create the motor. In practice, however, there is always a delay with the sampling time interval Tab. The circuit tends to oscillate if no special measures are taken.

Fig. 5 shows the step response of the current for a real drive. If the control deviation Δi (k) = isq_ref at t = kTab, the controller tries to output such a voltage usq_ref (k) that the control deviation disappears in the next control interval. Because of the delay of the modulator 3 , this voltage only arrives at the motor in the interval k + 1. As a result, the control deviation Δi (k + 1) remains unchanged, which results in the voltage setpoint or control output value: usq_ref (k + 1) = usq_ref (k). This results in a 100% overshoot of the current. With the training described below according to FIGS. 6 and 7 an attempt is made to solve this problem.

Looking at the stator equation in the q direction (this also applies to the d direction)

it can be determined that the current change can be predicted in the next control interval. From ( 4 ) the current prediction results in the difference equation:

The same effect occurs when usq_emk (..., K-2, k-1,..., K + 1, k + 2,...) Is used would be.  

Fig. 6 shows the achievable under combination with Stroprädiktion current waveforms. The predicted current already reaches the current setpoint isq_re after a sampling time interval Tab, with control deviation Δi (k + 1) equal to zero and the steady state of the current having been reached. The complete structure of the PI controller according to the invention in combination with current prediction is shown in FIG. 7.

According to FIG. 7, the control circuit indicated in FIG. 4 is expanded by a current prediction device 23 . This has a summing element 24 on the input side, to which the final, possibly limited controller output value usq_ref is fed at the first input. At the second input, the pilot control output value USQ_EMK is made available with a negative sign (in contrast to the summing element 21 of the I control channel). The first summing element 24 of the current prediction device 23 is immediately followed by a dead time element 25 , the time constant of which coincides with the sampling time interval Tab. The result of taking place in the summing junction 24 subtracting the pilot output value USQ_EMK from the final control output value usq_ref is thus a sampling Tab delayed to a further summing junction 26 is supplied, which is applied directly at the output of dead time 25 with its first, non-inverting input. At the second, inverting input of the summing element 26 , the result of mathematical relationships is supplied via stator parameters and currents, as can be seen in equations (4) and (5) above. The output of the summing element 26 is fed to a proportional element 27 , which effects a link with the parameter Tab / I _S (I _S : stator inductance of the synchronous machine). The output value Δpred_isq, which represents a predictive change in the actual current value, is then at the output of the current prediction device 23 . This is a further summing element 28-current value summed with the stator isq, and the result as a predictive current value pred_isq the above-mentioned comparison member 5 at its inverting input entered.

The accuracy of the current prediction according to equation (5) depends on machine parameters. This has a disadvantageous effect on the control accuracy of the current in the steady state. In order to eliminate this negative influence, the current prediction device 23 is switched off by means of a switch 29 when the control deviation of the current falls below a certain limit value ε.

The step response of the current with the controller according to FIG. 7 at a speed of 500 rpm is shown in FIG. 8. The current reaches the setpoint in approx. 250 µs without overshoot. This time corresponds to the desired minimum value, which can be calculated using equations (6) as follows:

FIG. 9 shows the step response of the current with a previously known PI controller, in particular according to FIG. 1. The controller parameters and the speed are set in accordance with FIG. 8. Although the increase in current here is exactly the same as with the regulator according to the invention, the approx. 40% overshoot is often unusable in practice. You have to set the PI controller slowly. FIG. 10 shows the step response of the current with a previously known PI controller, set according to the optimum amount, also at 500 rpm. It can clearly be seen that the current control loop has become much slower. The rise time is more than 750 µs.

FIG. 11 shows the step response for the controller according to the invention according to FIG. 7, and for comparison in FIG. 12 that for the previously known PI controller in each case at a speed of 2500 rpm. In comparison with FIG. 8 and FIG. 9, the rise of the current is smaller because of the higher induced voltage.

Reference list

PT1 motor winding

1

Controlled system
r _p

Stator winding resistance

2nd

Summer
usq control voltage

3rd

Pulse width modulator
Dead time constant tab, sampling time interval

4

Tracer
isq actual stator current value

5

Comparator
Δi control deviation

6

PI controller

7

Proportional link

8th

Integrator
and_P_part P control channel output value
and_L_part I control channel output value

9

Controller output summer

10th

Feedforward control

11

,

12th

Multiplier
speed speed, speed
usq_from provisional controller output value
usq_ref final controller output value

13

Limitation
usq control voltage
USQ_EMK output value feedforward control

14

Dead time limb

15

Adder

16

Multiplier

17th

Fraction factor element

18th

Divider

19th

Switching device

19th

a switching element

20th

Summer

21

Summer

22

Summer

23

Electricity prediction facility

24th

Summer

25th

Dead time limb

26

Summer

27

Proportional link

28

Summer

29

Switch off

Claims (17)

1. Method for discrete-time control of the stator current of an electrical machine, an actual value (isq) of the stator current being recorded at discrete sampling times and compared with a current setpoint (isq_ref), and a control deviation (Δi) resulting from the setpoint / actual value comparison by means of at least an I control to a control output value (usq_ref) for a control voltage (usq) of the stator or on the machine, the control output value (usq_ref) delayed by a sampling time interval (Tab) limited by two successive sampling times into the control voltage (usq) of the machine, characterized in that within the framework of the I control, a more recent control deviation (Δi) and an older control deviation, which are at least one sampling interval (Tab) back, are averaged together, and a resulting mean value over time for the control output value (usq_ref ) is integrated. 2. The method according to claim 1, characterized in that for averaging the younger and older control deviation (Δi) are added together and then the addition result is halved ( 17 ) before further integration. 3. The method according to claim 1 or 2, characterized in that the Control deviation (Δi) before averaging with one or more I control parameters (kp / Tn) are multiplied, linked or processed. 4. The method according to any one of the preceding claims, characterized in that, depending on an amount of a control output value (usq_out), which is preferably determined provisionally using the I control, the control deviation (Δi) before averaging an amount or size reduction ( 16 , 18 , 19 , 20 ) is subjected. 5. The method according to claim 4, wherein the I control is combined with a P control, characterized in that for limiting the amount ( 16 , 18 , 19 , 20 ) when a predetermined limit value is exceeded by the provisional control output value (usq_out), the control deviation ( Δi) is multiplied by a factor which is formed by dividing ( 18 ) the limit value (usq_lim) reduced by a result of the I control (u_I_Teil, USQ_EMK) by the result of the P control (u_P_Teil). 6. Method for controlling the stator current of an electrical machine, wherein an actual value (isq) of the stator current is detected and compared with a current setpoint (isq_ref), and a control deviation (Δi) resulting from the setpoint / actual value comparison by means of a combined P and I -Control to a control output value (usq_ref) for a control voltage (usq) of the stator is implemented, in particular according to one of the preceding claims, characterized in that a predictive current actual value (pred_isq) is used for the setpoint / actual value comparison ( 5 ) Addition ( 28 ) of the current actual value (isq) detected on the stator is formed with a current actual value change (Δpred_isq) determined by current prediction ( 23 ). 7. The method according to claim 6, wherein in the context of a time-discrete control in discrete sampling times, the actual value (isq) of the stator current is recorded, and the control output value (usqsef) directly calculated from the resulting control deviation (Δi) by a sampling interval limited by two successive sampling times ( Tab) is delayed ( 3 ) converted into the control voltage (usq), characterized in that a stator current change (Δpred_isq) is predicted as part of the current prediction ( 23 ) and with the last sampled current actual value (isq) for the subordinate setpoint / actual value comparison ( 5 ) is added ( 28 ). 8. The method according to claim 7, characterized in that for the addition ( 28 ) the current actual value (isq) detected at the current sampling time on the stator, and for the current prediction ( 23 ) a calculated in a sampling time interval (Tab) before the current sampling time Control output value (usq_ref) can be used. 9. The method according to claim 6, 7 or 8, characterized in that the current prediction ( 23 ) is switched off ( 29 ) when the control deviation (Δi) falls below a predetermined limit (ε). 10. I control channel ( 7 , 8 ) for a PI, PID or other controller for use in a current control loop of the stator winding of an electrical machine, in particular for carrying out the method according to one of the preceding claims, characterized in that in the I control channel ( 7 , 8 ) in front of an integrating element ( 8 ) there is a device ( 14 , 15 , 17 ) for forming an average value from a more recent and at least one older control deviation (Δi). 11. I control channel ( 7 , 8 ) according to claim 10, characterized in that the mean value generator ( 14 , 15 , 17 ) is a combination of a dead time element ( 14 ), an adder ( 15 ) downstream of this and a fractional factor element downstream of this ( 17 ), and a first input of the adder ( 15 ) and the input of the dead time element ( 14 ) in parallel with a proportional element ( 7 ) associated with a control deviation (Δi) for control parameters (kp / Tn) or with a direct supply for the control deviation (Δi) are connected, and the second adder input is connected directly to the output of the dead time element ( 14 ), and the fraction factor element ( 17 ) is designed to multiply the adder output by a fraction factor, the reciprocal of which is the number of averages ( 14 , 15 , 17 ) used, individual control deviation values (Δi). 12. I control channel according to claim 11, characterized in that the dead time of the dead time element ( 14 ) is dimensioned after a sampling interval (Tab) of an external, responsive to discrete sampling times sampling element ( 4 ) for time-discrete detection of individual current actual values (isq), and input I-channel control (7, 8) being associated with the scanning member (4) with discrete actual current (i sq) fed setpoint / actual value comparison device (5). 13. PI controller for use in a current control loop of the stator winding of an electrical machine, with a P control channel (kp) and an I control channel ( 7 , 8 ) combined with this, which is designed according to one of the preceding claims, characterized in that that in the I control channel ( 7 , 8 ) the dead time element ( 14 ) and the adder ( 15 ) has a limiting device ( 16 , 18 , 19 , 20 ) for control deviations ( 16 , 18 , 19 , 20 ) activated depending on the size or the amount of a preferably preliminary controller output value (usq_out) Δi) or other input values. 14. PI controller according to claim 13, characterized in that the limiting device ( 16 , 18 , 19 , 20 ) has a multiplier ( 16 ) arranged upstream of the dead time element ( 14 ) and the adder ( 15 ), which has a control deviation (Δi) or multiplies the output value of a proportional element ( 7 ) processing the control deviation (iM) by the output value of a dividing element ( 18 ), which is reduced by a summing element ( 20 ) by an output value (u_I_Teil, USQ_EMK) of the I control channel ( 7 , 8 ) Limit value (usq_lim) as a dividend and is fed from the initial value (u_P_Teil) of the P control channel (kp) as a divisor. 15. Current control loop of the stator winding of an electrical machine, with a setpoint / actual value comparison element ( 5 ) fed with current setpoint and actual current values (isq_ref, pred_isq) and a PI controller ( 6 ) connected downstream thereof, in particular according to one of the preceding claims, characterized by a device ( 23 ) for current prediction with at least controller output values (usq_ref) and current actual values (isq, isd) and optionally an EMF pilot control value (USQ_EMK) as input variables and a current actual value change (Δpred_isq) as a resultant output variable, which are transmitted via a summing element ( 28 ) to the Current actual value (isq) linked and / or combined to the comparator ( 5 ) as a predictive current actual value (pred_isq) is supplied. 16. Current control circuit according to claim 15, which operates on the basis of a sampling element ( 4 ) for the actual stator value (isq) in a time-discrete manner, characterized in that the prediction device ( 23 ) is a dead time element ( 25 ) with a dead time constant corresponding to a sampling interval (Tab) of the sampling element ( 4 ) between two discrete sampling instants, and the dead time element ( 25 ) the PI controller output value (usq_ref) or a difference ( 24 ) between the PI controller output value (usq_ref) and the EMF pilot control value (USQ_EMK) is supplied, and the prediction device ( 23 ) furthermore one or more computing elements ( 26 , 27 ) for linking the output value of the dead time element ( 25 ) with the actual stator current value (isq, isd) and / or with machine parameters (r _S , I _S ) and possibly with the EMF pilot control value (USQ_EMK) for the purpose of determining the predictive current actual value change (Δpred_isq). 17. Current control circuit according to claim 15 or 16, characterized by a switch-off device ( 29 ) which can be activated by a threshold value decider for the prediction device ( 23 ), the threshold value decoder responding to falling below a predetermined threshold value ( 8 ) by the control deviation output by the comparison element ( 5 ) (Δi) is formed.