EP0349568A1

EP0349568A1 - Improved control method of an ac induction motor and device therefor

Info

Publication number: EP0349568A1
Application number: EP88902606A
Authority: EP
Inventors: Ragnar JÖNSSON
Original assignee: Individual
Current assignee: Individual
Priority date: 1987-03-11
Filing date: 1988-03-11
Publication date: 1990-01-10
Also published as: JPH02502510A; SE8701008D0; WO1988007287A1; AU1487788A; SE8701008L

Abstract

Le procédé et le dispositif décrits servent à commander un moteur à induction à courant alternatif à n'importe quelle vitesse du moteur y compris la vitesse zéro. La tension d'alimentation est modulée au moyen d'un signal de forme sinusoïdale avec une fréquence prédéterminée, de façon à produire un champ magnétique statorique et un courant rotorique. La tension d'alimentation (uS) subit une distortion, afin de conférer à la partie de magnétisation de la tension (VL) une forme sinusoïdale. Si cette condition est remplie, on obtient une compensation des erreurs dues par exemple aux propriétés de magnétisation non linéaire du fer au début de la saturation et les causes supplémentaires d'ondulation du couple moteur sont éliminées. La présente invention se rapporte à la commande en tension dudit moteur à induction malgré le fait que le couple soit produit par le courant traversant les enroulements. La distortion des tensions peut se produire de différentes manières et des tables de sinus et cosinus avec valeurs ajustées sont décrites. On donne ici une analyse mathématique complète du modèle dudit moteur à induction à courant alternatif, analyse qui comprend également les inductances de fuite au niveau du stator et du rotor.The method and the device described serve to control an AC induction motor at any speed of the motor including zero speed. The supply voltage is modulated by means of a signal of sinusoidal shape with a predetermined frequency, so as to produce a stator magnetic field and a rotor current. The supply voltage (uS) is distorted to give the voltage magnetization part (VL) a sinusoidal shape. If this condition is fulfilled, an error compensation is obtained due for example to the non-linear magnetization properties of the iron at the start of saturation and the additional causes of ripple of the motor torque are eliminated. The present invention relates to the voltage control of said induction motor despite the fact that the torque is produced by the current passing through the windings. Distortion of voltages can occur in different ways and sine and cosine tables with adjusted values are described. A complete mathematical analysis of the model of said alternating current induction motor is given here, an analysis which also includes the leakage inductances at the stator and rotor.

Description

IMPROVED CONTROL METHOD OF AN AC INDUCTION MOTOR AND DEVICE THEREFOR

FIELD OF INVENTION

The present invention relates to an improved method for voltage control of an AC induction motor. The invention also relates to an electronic circuit for performing said method. The invention is based on the disclosure of Swedish Patent No. 8000118-3 (equivalent to US-A-4,458,193) which is incorporated herein by reference.

The Swedish Patent is based on an idealized motor model with linear magnetic properties and neglectable leakage inductances. The present invention is an improvement thereof. It eliminates the torque ripple caused by the nonlinear iron magnetization curve, especially at low speed. The leakage inductances have been added to the motor model with the corresponding additions to the control system. Thus, two major differences between the idealized motor model for the control system and the real motor have been eliminated. PRIOR ART

It is easy to control the AC induction motor speed by varying the voltage and frequency, or current and frequency. A great number of such systems already exist on the market for simple applications, such as pump and fan drives etc. The present invention relates to more advanced systems, which make it possible to control the AC induction motor in all kinds of drives, also in servo drives having fast dynamic response to control signals. In the following primarily such advanced drives are considered. The AC induction motor has a very complicated mathematical model. In modern literature, for example the book "Control of Electric Drives" by professor Werner Leonhard (Springer Verlag, 1985), a theoretical motor model, "tailored to the needs of controlled drives", is given. The voltages and currents are described by complex two-dimensional vectors. The motor model results in one complex differential equation for the stator and one complex differential equation for the rotor, plus one complex differential equation for the torque calculation. No analytical solution of the non-linear equations exists, according to professor Leonhard.

The mathematical treatment, is different in the case of voltage control and current control of the induction motor. Current control is normally preferred. However, there are some problems associated with current control.

Current control systems use current generators forcing currents into the stator windings, whereby the stator differential equation is eliminated which means a very significant simplification of the mathematical problem.

There are a number of different methods for current control of the AC induction motor. One method, as described by professor Leonhard in the above-mentioned book, uses a 16-bit microprocessor in a control system based on the rotor equation together with the concept of "field oriented control". Other methods, described in a number of patents, are based on "slip theory" together with "vector control".

Some of the problems associated with current control may be summarized: 1 ) The control problem has not yet been completely solved in a satisfactory way.

2) If the current generator saturates or reaches the "ceiling voltage" (the supply voltage), it no longer operates as a current generator and the whole control system may be brought out of balance. 3) A current generator must have a high bandwith, significantly higher than all other parts of the system resulting in expensive power circuits.

4) Since the stator equation is eliminated, the control system is depending on the rotor equation, and is thus critically dependant on the rotor inductive time constant. This time constant varies with rotor resistance, which varies with rotor temperature, and it is difficult to measure the rotor temperature. However, this problem can be solved. A more serious problem occurs when the rotor iron goes into magnetic saturation and the rotor time constant collapses. Then the whole control system may collapse.

5) Exactly as in the DC motor, a counter electromotive force "counter emf" is generated in the rotor winding in proportion to the rotor speed. This counter emf has the same effect as a tachometer feedback signal, creating the necessary damping of outer control loops and a separate tachometer is often not necessary. When a current generator is used, it eliminates the effect of the counter emf, and thus a separate tachometer function is necessary.

6) There is a nonlinear relationship between current and magnetic field strength. This πonlinearity is a source of torque ripple.

However, there is one significant advantage with current control. From DC motor systems it is well known that the current generator is preferred in high response servo systems, because the current generator forces current through the leakage inductances of the motor, thus eliminating or reducing the inductive time constant. Most often it is this inductive time constant that limits the obtainable system bandwith. The same is true for AC induction motor systems. However, motor control systems can be designed without current control, and there are other methods to improve the servo bandwidth, for example lead-lag networks in the signal circuit. Voltage control is the natural solution, and in spite of the fact that motor torque is generated by currents, there are strong reasons to use voltage control, as shown below.

The Swedish Patent No. 8000118-3 (US-A-4,458, 193) discloses an AC induction motor control system based on the use of a voltage control vector, composed of two orthogonal signals. This control vector is modulated with the sine and cosine of a modulating frequency before connection to the motor stator windings. This system is based on a simplified and linear model of the AC induction motor and is simple and fulfils almost all essential demands on such a system.

When a standard, currently available AC induction motor is connected to voltage generators with perfect three-phase sine waves of variable amplitude and frequency, a certain torque ripple can be observed especially at low speeds, approximately 5 - 10 % of the actual torque. Especially at zero speed, it is easy to measure the torque. There are several possible sources of this torque ripple, for example:

1) Differences between the real motor design with a limited number of winding slots and the theoretically optimal motor designwith continuous winding distribution.

2) Unsymmetrical stator winding design and manufacture.

3) Eccentric rotor location in the airgap.

4) Oval stator.

5) Bearing play. 6) Inhomogenities in the current conductors of the squirrel cage rotor.

7) Unsymmetrical magnetic properties in the stator and rotor laminations.

8) Iron magnetic saturation. Some of these ripple sources can be compensated by individual trimming of the individual phase signals, but this is not the optimal solution. It is better to eliminate the ripple sources by careful design of the motor.

However, in order to make efficient use of the magnetic material, the iron has to be magnetized into the beginning of the saturation region. This is an inherent source of torque ripple which cannot be eliminated by the design of the motor. Thus, it is important to design a motor control system that compensates for the torque ripple due to magnetic saturation. It is not necessary to eliminate the ripple completely. For comparison, it should be noted that also DC motors have a certain amount of torque ripple.

The non-linear magnetization curve of the iron is an unwanted error source in the motor. So are the leakage inductances. They represent approximately 5 % of the total inductance. They are parasitic elements without any contribution to the torque generation. They should be made as small as possible in modern motor designs. However, since they cannot be completely eliminated, it is important to include them in the motor model and in the control system design. DISCLOSURE OF THE INVENTION It is an object of the present invention to provide a method of controlling an AC induction motor for controlling or compensating the feed voltage so that the magnetizing part of said voltage is sine shaped.

Another object of this invention is to provide a method of controlling an AC induction motor for compensating for torque ripple due to a nonlinear magnetization curve of the iron. A further object of this invention is to disclose a control method and an electronic circuit that takes into account the leakage inductances of the stator and the rotor.

This invention may be used together with the Swedish Patent No. 8000118-3 (US-A-4,458,193) mentioned above, and together with all similar control systems for the AC induction motor.

The invention will appear from the appended claims. The particulars of the invention will appear from the detailed description below of the theory underlying the present invention and the description of several embodiments with reference to the appended drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 is a schematic diagram of a two-phase induction motor according to Fig. 3 of the above-mentioned Swedish Patent No. 8000118-3. Fig. 2 is a schematic diagram of the equivalent circuit of one phase in the induction motor, according to Fig. 4 of the above-mentioned Swedish Patent No. 8000118-3.

Fig. 3 is a phasor diagram showing voltages and currents in one phase of the induction motor, according to Fig. 5 of the above-mentioned Swedish Patent No. 8000118-3.

Fig. 4 is a block diagram according to Fig. 8 of the above-mentioned Swedish Patent No. 8000118-3.

Fig. 5 is a typical magnetization curve for iron. Fig. 6 is a block diagram of an apparatus for control of an induction motor, including compensation for the nonlinear iron magnetization curve, according to the invention.

Fig. 7 is a block diagram similar to Fig. 6 but including another compensation circuit.

Fig. 8 is a schematic diagram of a two-phase resolver. Fig. 9 is a schematic diagram of a three-phase resolver.

Fig. 10 is a block diagram similar to Fig. 6 comprising a three-phase resolver. Fig. 11 is a coordinate transformation scheme. Fig. 12 is a phasor diagram similar to Fig. 3 comprising the complete model of the induction motor.

Fig. 13 is a schematic diagram similar to Fig. 2 of the equivalent circuit of one phase of the induction motor.

Fig. 14 is a block diagram similar to Fig. 4 of the complete circuit.

Fig. 15 is a schematic diagram of the magnetizing current distorted according to the invention. DETAILED DESCRIPTION OF THE THEORY AND OF THE EMBODIMENTS

All equations and most of the figures relate to a two-phase, two-pole motor, because the theory is based on a two-phase system with 90 geometric difference between the two phases. The generation of torque depends on interaction between two orthogonal phases. Fig. 1 is taken from the above-mentioned Swedish Patent No. 8000118-3, and shows supply voltages V₁, V₂ and generated magnetic field components B₁, B₂. Only those rotor windings which have maximum coupling to the stator windings are shown. I₁, I₂ are the induced currents in the rotor winding . The relations between the above-mentioned quantities are given below:

V₁ = V₀ sin wt B₁ = -B₀ cos wt I₁ = I₀ sin wt

V₂ = V₀ cos wt B₂ = B₀ sin wt I₂ = I₀ cos wt

w = signal frequency (radians/second)

The magnetic field from one phase cooperates with the rotor current from the other phase. This is of fundamental importance for the operation of the motor. Thus, current I₁, generates torque together with magnetic field B». In the same way, current I₂ generates torque together with magnetic field B₁. As can be seen from the formulas above, the cooperating quantities have the same phase angle which is necessary for maximum torque generation. Thus the torque is proportional to: I₁ x B₂ + I₂ x B₁ = = I₀ sin wt x B₀ sin wt + I ₀ cos wt x B₀ cos wt = = I₀ x B₀ x ( sin² wt + cos² wt ) =

= I ₀ x B₀

There is no time dependant torque ripple. It is, however, necessary that both the magnetic field strengths B₁, B₂ as well as the rotor currents I₁, I₂ are sinusoidal. Normally, the motor is operated with constant field strength B₀ and variable rotor current I₀. Then the torque is proportional to I₀.

It is important to understand, that the constant torque is obtained as a result of the perfect interaction in time between the sine and cosine quantities in the formula. Thus it is the sinusoidal form in time that is important, not the sinusoidal form in the space. Also Figs. 2 and 3 are taken from the above-mentioned Swedish Patent No. 8000118-3. Fig. 2 is a schematic diagram of the equivalent circuit of one phase of the induction motor. The circuit includes only the primary components of the theoretical model. The broken lines represent the air-gap between stator and rotor. Components to the left of the broken lines represent the stator, and components to the right represent the rotor. Stator resistance is R_S, stator inductance is L₀ and rotor resistance is R_R. Supply voltage is u_S. Total current on the stator side is i_S . Inductive stator current is i_M and rotor current is i_R. The magnetizing voltage V_L is transformed from the stator to the rotor through the airgap. A counter-electromotive voltage U is induced in the rotor winding. U is proportional to rotor speed. The transformation ratio between the stator and rotor windings is assumed to be "1". Any real motor parameters can be converted to this transformation ratio. The stator and rotor leakage inductances are neglected.

Fig. 3 is a phasor diagram showing voltages in one phase of the induction motor. The inductive current i_M is used as a reference phasor, pointing to the right in the diagram. The resistive current in is 90 ahead, pointing upwards in the diagram. Fig. 4 is a block diagram of an electronic circuit constructed according to the phasor diagram of Fig. 3 and according to Swedish Patent No. 8000118-3.

The total supply voltage vector u_S to the motor is composed of two orthogonal components S₁ and S₂, where S₂ is substantially a constant voltage, compensating for the resistive stator voltage drop caused by the inductive stator current and S₁ is the sum of the magnetizing voltage V_L plus the resistive stator voltage drop caused by the rotor current. If any of these components (S₁ and S₂) are incorrect, the total supply voltage vector u_S will be affected and hence all parts of the system. Thus, it is important that both components are correct.

Both the magnetic field strength B and the rotor current i_R must be modulated with undistorted sine and cosine waveforms. Both these quantities depend directly on the magnetizing voltage V_L. Thus, the magnetizing voltage V_L is of vital importance for the motor function, since it is the common supply voltage for the stator inductance as well as for the rotor circuit. The rotor current i_R is resistive, and if V_L is a sine signal, the rotor current will also be a sine signal. The application of a well-known electromagnetic law reveals an interesting fact about the stator magnetic field, and this is the very heart of the present invention. The magnetizing voltage V_L is equal to the time derivative of the total magnetic flux linkage, and thus of the total magnetic field strength B. Thus:

V_L = (dB / dt) x C where C is a constant

Thus, if V_L is a sine signal, B will be a cosine signal, but the inductive (magnetizing) current i_M may have a different waveform. Fig. 5 shows a typical magnetization curve for iron (the hysteresis effect has been neglected) and if the magnetic field strength B is known, the corresponding inductive current _iM can be obtained. The curve is not linear, and at higher field strength, the current increases faster than the field strength.

Thus, it has been shown that the magnetic field should be controlled by a voltage (V_L) although it is created by a current.

Fig. 6 is a block diagram of a preferred embodiment of a complete circuit for voltage control of an AC induction motor, with compensation for the nonlinear magnetizing current according to the present invention. This block diagram is identical to that in the Swedish Patent No. 8000118-3 (US-A-4,458,193), except for the nonlinear compensation. Block 5 in the Swedish Patent has been replaced by the blocks 6 - 10 in Fig. 6.

An external control voltage vector S₁, S₂ is generated according to the phasor diagram. This is a stationary vector (not rotating). In the circuit, the vector is modulated by sine and cosine signals in four multipliers 1. The four multiplier outputs are added respectively subtracted in the adders 2. The four multipliers and the two adders together form a so called "resolver" or vector rotator. The rotating output vector is then fed to the power circuits 3, which control the motor 4.

The modulating frequency w is calculated according to the Swedish Patent No. 8000118-3 (US-A-4,458,193), or any other suitable method. Block 6 is a so called "modulo-2π" integrator, which integrates the input frequency w and delivers the output angle. At the angle (2π) the integration resumes from zero. Integration can be performed forward and backwards. Block 7 is a sine table, delivering the sine of the input angle to the multiplier 1. Block 8 is a cosine table. Block 9 is a sine table, modified according to the desired waveform of the magnetizing current. Block 10 is a cosine table, modified according to the desired waveform of the magnetizing current.

Such a "modulo-2π" integrator can easily be made as a counter in digital technique. The counter counts incoming pulses up to its upper limit, which corresponds to the angle 2π, and the next pulse will overflow the counter, which now shows zero. The output signal from the counter will be an angle which is input to the tables 7 - 10. According to the Swedish Patent, the modulating frequency w is obtained by addition of two different signals, one proportional to the motor speed (or counter emf) and one proportional to the signal S₁. If an incremental position encoder is connected to the motor shaft, the pulse train from this encoder has a frequency which is proportional to the motor speed. This pulse train can be used as input signal to the "modulo-2π" counter in block 6. It is convenient to make also signal S₁ as a pulse train, which can be added to the pulse train from the position encoder and integrated in the counter. It is observed that this principle is very easy to implement in a microprocessor-controlled system. Also, if the control system is integrated into an "application specific integrated circuit" ASIC, it is preferred to use this "digital" approach.

Moreover, it is convenient to observe that a good motor design should have approximately the same stator and rotor resistance (as seen on the stator side). In that case the contribution of the two signals to the counter is approximately equal, which gives an indication on how to scale the two signals for obtaining the correct frequency w.

In order to illustrate the operation of the invention, below some data pertaining to a working model is given. Thus, the voltage S₂ can be about 20V, while the voltage S₁ can vary between +380V and -380V. When the motor is standing still, S₁ will be zero and the component S₂ will dominate, but at medium speed and maximum speed, S₁ dominates. Thus, both components are important, each one in its specific speed area (S₁ at high and medium speeds and S₂ at low speeds).

Thus, the improvement according to the present invention will be most important at low speeds and zero speed.

The modulation with sine and cosine waveforms does not mean that the waveforms to the motor always are sine shaped. The control signal S₁ comes from the outer control loop, and may change very quickly. This has direct influence on the motor voltages.

Of course it is possible to generate the control signals in other ways, especially if a microprocessor is used for the calculations. Fig. 7 shows how the invention can be adapted to different magnetization fields B₀. The blocks 9 and 10 of Fig. 6 are replaced by two blocks 11 in Fig. 7. Block 11 comprises a sine or cosine table 12, the input signal of which being the control signalα. The output of block 12 is fed to a multiplier 13 for multiplication with a constant B₀. The output from block 13 is fed to another table comprising the nonlinear magnetization curve of the iron and the output from block 14 will be the magnetizing current i_M. It is also shown in Fig. 7 that the resistance R_S may be adjusted by adding a signal C x R_S x T to the value of R_S (C = a constant and T = temperature). Said signal may be obtained from a temperature sensor located on the stator.

As an alternative, it is possible to compensate for different values of the magnetic field B₀ by having several tables in blocks 7 to 10, one for each discrete value of the magnetic field B₀.

When B₀ is changed, corresponding parameters in other parts of the control system have to be changed.

It should be mentioned that the tables in blocks 9 and 10 should take into account that the magnetic field passes iron as well as air. In practice, tables may be created by experiments. In Fig. 15, there is shown an example on the waveform for a typical magnetizing current

The description up to now has been based on a simplified model of the motor essentially according to the above-mentioned Swedish Patent

No 8000118-3 (US-A-4,458,193) and has added compensation for the nonlinear magnetization curve of the iron. It is possible to extend the motor model to include the leakage inductance of the stator and the rotor. In Fig. 13, the completed equivalent circuit of one phase of the induction motor is shown. The leakage inductances for the stator and the rotor have been added. Note that the direction of the rotor current i_R is the opposite in relation to Fig. 2.

A more exact mathematical description is given in the equation's enclosure, equation 1 and 2. These equations are the same as given in the above-mentioned book "Control of Electric Drives" by professor Werner Leonhard, chapter 10, equations 10.38 and 10.39. These equations are the starting point for the following analysis. The calculations are based on stator coordinates, rotor coordinates and field coordinates. The relationship between these coordinates appears from Fig. 11.

Equation 3 defines the magnetizing current. In equations 4, 5 and 6, the currents i_M, i_R and i_S are converted to field coordinates and inserted in equations 1 and 2 resulting in equations 8 and 9, where the stator current is eliminated. These equations are differentiated resulting in equations 10 and 11, and converted to field coordinates in equation 13 and 14. Equations 13 and 14 are exact and are the basis for the following synthesis.

The motor is controlled by the voltage vector S, which is field oriented. The control vector S is defined in equation 12 and is calculated mathematically exact as shown below. The only condition is that the rotor flux should be constant. In practice even this condition is not absolutely necessary. It is well-known that unwanted crosscoupling is eliminated by having the rotor flux constant.

Equations 15 - 18 define the magnetic flux. Equations 19 and 20 are obtained from equation 18. Equation 20 is true only if the rotor flux is constant. Equations 15 - 20 are inserted in equations 13 and 14 resulting in equations 21 and 22. Equation 23 is obtained from equation 22 and the Laplace operator s is introduced in equation 24. The stator equation 25 is obtained from equations 21 - 24. The present control system is based on stator equation 25. Equation 26 defines the rotor flux as reference and equation 27 defines the control vector S. The stator equation 25 is split into its real and imaginary parts by means of equations 26 and 27 resulting in equations 28 and 29. The rotor current i_Rf is defined in equation 30 according to equation 23. Equations 31 and 32 are the same as equations 28 and 29 but inserting the rotor current. Equation 33 calculates the field angle frequency from equation 29.

Equations 31 and 32 are shown graphically in Fig. 12. The influence of the added leakage inductances are shown in the extended block diagram of an apparatus for control of an induction motor shown in Fig. 14 which is based on equations 28 and 33. This block diagram is essentially the same as shown in Fig. 4, but the scaling factors for frequency calculation have been extended to include the leakage inductances. However, a new term has been added to the control voltage S₂ as shown in Fig. 14. The stator frequency w is multiplied in block 21 with the difference between the stator frequency and rotor speed from block 22. The result is multiplied in block 23 with a constant and subtracted in block 24 from the original S₂ signal.

These equations seem to be complex but are rather easy to use in a control system. In practice it will not be necessary to exactly follow the theoretical results but a course dimensioning from the formulas given can take place and then be adjusted according to the results. It can also be preferred to use standardized control systems standardizing certain parameters of the control system. The mathematical motor model is valid for steady state conditions as well as for transient conditions. The equivalent circuit in Fig. 13 helps to give a general understanding. Transient signals are transformed to the rotor essentially without changing the stator magnetic field, and thus the transient response, including the effect of the stator and rotor leakage inductance can be estimated. Consequently, the control system based on this motor model is useful for steady state conditions as well as transient conditions.

Above, the invention has been described in relation to a two- phase motor with two poles. Of course motors with more than two poles can be used with this invention. Most motors are available as three- phase motors. Normally it is very easy to transform the sinusoidal signals from two-phase to three-phase by simple addition and subtraction. However, with the present system, the signal S₂ is no longer sinusoidal and simple addition is no longer permitted. Thus, it will be necessary. to generate the signals for each phase separately as shown in Fig. 9 and 10. Fig. 8 shows a resolver symbol for a two-phase resolver according to Fig. 6. Fig. 9 shows a resolver symbol for a three-phase resolver including the angular relationship between the output signals.

In Fig. 10 the components 1, 2 and 3 of Fig. 6 have been replaced by six multipliers 1', three adders 2' and three power circuits 3'. There are three blocks 15, 16 and 17 generating sinrx, sin (α - 120) and sin (α - 240) and three blocks 18, 19 and 20 generating modified cosoc, modified cos (α- 120) and modified cos (α- 240). It is clear that each resolver output is fed to each winding through the corresponding power amplifier 3'. Further sources of errors can be included in the motor model and in the control system. Thus, the hysteresis effect of the iron magnetization can be added to block 14 in Fig. 7. However, the hysteresis effect is quite complicated and not easily described. The use of soft iron with small and neglectable hysteresis is the best solution for motor and control system design.

The present invention describes how the control system takes care of two different unwanted phenomena in the AC induction motor, namely the nonlinear iron magnetization curve and the leakage inductances. The two phenomena and their influences on the control system have been described separately, but of course it should be possible to include them in a common control system. Compensation for the nonlinear magnetization current was very easy to achieve, because the total signal S₂ was proportional to the nonlinear magnetizing current. If the leakage inductances are added to the motor model and to the control system, there will be a new component in S₂, proportional to the rotor current. Then the original S₂ component has to be modulated by a distorted sine wave and the new S₂ component has to be modulated by a normal sine wave. However, this is no problem in a modern digital system, but in practice one may prefer to simplify the system, depending on actual requirements and actual motor data.

Finally, a few remarks about motor design. Classical as well as modern motor theory is based on the concept of a rotating magnetic field. Thus, motors are designed with a quasi-continuous sinusoidal distribution of the stator windings, in order to obtain a spatial sinusoidal distribution of the magneto-motive force in the airgap, cf for example the book "Control of Electric Drives" page 146. But, according to the explanation of Fig. 1, the sinusoidal distribution in time is enough to guarantee a constant torque generation, without ripple. The stator magnetic field is described as the sum of two individual, orthogonal, pulsating magnetic fields, superimposed on each other. Possibly, the spatial sinusoidal distribution of the stator winding is unnecessary. Then a new motor design, with only two phases, and with each pole encircled by the whole phase winding, would permit a more efficient use of the material, both copper and iron.

The AC induction motor is the most widely used motor today. The industrial standard motor is an excellent motor, and it is a great step forward that this motor now can be used with variable speed. There is a great market for standard motors in simple variable speed drives and medium performance servo drives. However, in high performance servo drives, the standard motor has to be improved in order to reduce the different sources of torque ripple, and only then the full value of this invention can be realized.

The description of this invention has shown a surprising result. A simple physical law proves that the AC induction motor should be voltage controlled instead of current controlled. Voltage control automatically eliminates the effect of the nonlinear iron magnetization curve. Only a small correction term has to be modulated according to the stator voltage drop caused by the nonlinear magnetization current. This correction term is most important at low speed operation of the motor, where S₂ dominates. At higher speeds, the control vector component S₁ dominates because it is much greater than S₂. It is possible still within the scope of the invention, to make S₂ larger than the nominal value, in order to get a stronger magnetic field at zero speed and low speed. The modulation of S₂ must be adjusted to the actual shape of the magnetizing current.

The AC induction motor is originally designed for operation on the sine shaped 50 Hz or 60 Hz line voltage. Because of the nonlinear iron magnetization curve, the motor current is a slightly distorted sine shaped current. However, the motor torque is constant without ripple. Current control would generate a torque ripple.

A further advantage of voltage control is that it automatically compensates for certain errors. As long as the magnetizing voltage V_L is correct, the magnetic field B will be correct independent of errors like air-gap variations etc.

Thus, it has been shown, that the AC induction motor is originally and inherently a voltage control motor, and should be used as such. The control system according to the invention is useful not only for speed control, but also for torque control and position control, depending on the requirements.

The invention is only limited by the appended claims.

DEFINITIONS

Stator resistance R_S

Rotor resistance R_R

Main inductance L _O

Stator leakage inductance l_S

Rotor leakage inductance l_R

(L_S = L_O + l_S L_R = L_O + l_R)

Magnetic field angle (rotor field relative to the stator) α

Stator frequency

Rotor angle (relative to the stator) ∈

Rotor speed

Slip angle α-∈

Slip frequency ω-∈

Vector rotation in positive direction e^Jα= cosα + j sinα

(Positive rotation = counter-clockwise)

Stator current vector (stator coordinates) _iS

Magnetizing current vector (stator coordinates) i_M

Stator voltage vector (stator coordinates) u_S

Rotor current vector (rotor coordinates) i_R

Magnetizing current vector (field coordinates) i_Mf

Rotor current vector (field coordinates) i_Rf

Control voltage vector (field coordinates) S

( S = S₂ + j S₁ )

Airgap flux (stator coordinates) ΔO

Rotor flux (stator coordinates) ΔR

Claims

PATENT CLAIMS

1. A method of controlling an AC induction motor, having stator windings and a rotor, for controlling the torque at all speeds of the motor including zero speed, comprising supplying a feed voltage (u_S) to the stator windings, said voltage being modulated by a substantially sine shaped signal having a predetermined frequency for providing a magnetic field by the stator windings and a current in the rotor winding, c h a r a c t e r i z e d by controlling or compensating the feed voltage (u_S) so that the magnetizing part of said voltage is sine shaped.

2. A method according to claim 1, said feed voltage (u_S) being composed of a magnetizing voltage (V_L) and a voltage drop in the resistive part of the stator winding, c h a r a c t e r i z e d by compensating for the nonlinear magnetization properties of the iron core of the motor by distorting the feed voltage so that the magnetizing voltage (V_L) is sine shaped.

3. A method according to claim 2, said voltage drop in the resistive part of the stator winding being composed of an inductive part and a resistive part, c h a r a c t e r i z e d by distorting said inductive part.

4. A method according to claim 1, whereby two electric control signals (S₁, S₂) are provided for controlling the induction motor, said control signals being connected to a resolver multiplying said control signals with a rotating vector for providing rotating output voltage signals (V₁, V₂), the rotation speed of the rotating vector being determined by a control system and being calculated by multiplying the first control signal (S₁) by a constant, and multiplying a rotor speed signal by a second constant, and adding the two signals for providing a stator frequency control signal, which is integrated for providing said rotating vector, c h a r a c t e r i z e d by subtracting the stator frequency control signal and the rotor speed signal and multiplying the difference signal with the stator frequency control signal and a scaling constant and subtracting the resulting signal from a second constant for providing said second control signal (S₂).

5. A device for performing the method according to claim 1, for controlling an AC induction motor comprising stator windings and a rotor, for controlling the torque at all speeds of the motor including zero speed, whereby a feed voltage u_S. is supplied to the stator windings, said voltage being modulated by a resolver with a substantially sine shaped signal having a predetermined frequency for providing a magnetic field by the stator windings and a current in the rotor winding, c h a r a c t e r i z e d by a compensation circuit (9, 10) for compensating or distorting the modulation signals supplied to the resolver (1, 2) so that the magnetizing voltage is sine shaped.

6. A device according to claim 5, said resolver being fed with two input signals, the first signal (S₁) of which being modulated with a sine signal and the second signal (S₂) of which being modulated with a cosine signal for providing a composite signal constituting said feed voltage (u_S), c h a r a c t e r i z e d by said compensation circuit (9, 10) being adapted to influence upon said second signal (S₂) for providing a sine shaped magnetizing voltage.

7. A device according to claim 5 or 6, c h a r a c t e r i z e d in that said compensation circuit (9, 10) is a sine and cosine table in which the values have been adjusted for said compensation.

8. A device according to claim 5, c h a r a c t e r i z e d in that said AC induction motor has three stator windings and that said resolver is a three-phase resolver generating three signals having a phase angle of 120 in relation to each other.

9. A device according to anyone of claims 5 - 8, c h a r a c t e r i z e d by including the leakage inductances of the rotor and stator in said compensation by including a correction term for said second signal.

10. A device according to claim 9, whereby two electric control signals (S₁, S₂) are provided for controlling the induction motor, said control signals being connected to a resolver multiplying said control signals with a rotating vector for providing rotating output voltage signals (V₁, V₂), the rotation speed of said rotating vector being determined by a control system and being calculated by multiplying the first control signal (S₁) by a constant in a first block, and multiplying a rotor speed signal, obtained from a tachometer, by a second constant, in a second block, adding the two signals for providing a stator frequency control signal, which is integrated for providing said rotating vector, c h a r a c t e r i z e d by an adder for subtracting the stator frequency control signal and the rotor speed signal and a multiplier for multiplying the difference signal with the stator frequency control signal and a scaling constant and an adder for subtracting the resulting signal from a constant for providing said second control signal (S₂).