REGULATION OF AND REGULATOR FOR A PERMANENT MAGNET SYNCHRONOUS MOTOR
The present invention relates to the regulation of a permanent magnet synchronous motor and to a regulator for one.
The regulation of a permanent magnet motor is often problematic in the most exacting application solutions. A particularly exacting application solution is the elevator environment. More generally, it can be said that the application solution is particularly exacting if it involves extraordinarily accurate position, speed or torque regulation or if one or more of these aspects of regulation requires extraordinary accuracy.
Usually an exacting application solution contains a plurality of cascaded and/or parallel time constants in the control loop that influence the regulation, which is a source of instability in regulation.
Another problem associated with che regulation of an permanent magnet synchronous motor is the lack of directly measurable indicator signals. As such signals are needed in the feedback, the generation of adequate signals requires relatively heavy computation. However, such computation creates a delay that impairs the stability of regulation.
En example of prior art is presented in ANALOG DEVICES data sheet "AC Vector Processor AD2S100".
The object of the present invention is to eliminate the drawbacks of prior-art technology -and especially to achieve a new method for the regulation of a permanent magnet synchronous motor that is more immune to interference and more stable than earlier regulation systems .
The features characteristic of the invention are presented m the claims .
The invention makes it possible to achieve a particu- larly stable regulation system for a permanent magnet synchronous motor. The regulation system is especially applicable to the elevator environment. The invention provides a clear principle for the construction of a regulator for a synchronous motor implemented using per- manent magnets.
When the longitudinal current m a permanent magnet synchronous motor is adjusted to zero, the rest of the regulation and control of the motor becomes simpler. Adding the third harmonic of the control voltage to the result of a coordinate system transformation of the control voltages makes it possible to use a smaller inverter. The number of poles m the motor can be readily taken into account m the coordinate system transforma- tions. This allows regulation that accurately and quickly follows the motion of the motor. Such a motor is especially applicable as an elevator drive motor with feedback effected via a signal proportional to tne elevator speed or to the traction sheave position. A pref- erable solution in the regulator is to form a 3 -phase reference for the motor supply voltage from the uq and ud voltages by transforming them to produce components corresponding to each phase and by combining the components for each phase, thus reducing the computation work or allowing the use of constant transformations, which can be implemented using standard components . The required computation work can also be reduced by using memory components instead of computation.
In the following, the invention will be described by the aid of a few examples of its embodiments without limit-
ing its sphere of application, by referring to the at¬ tached drawings, wherein
Fig. 1 presents a block diagram representing a regu¬ lator for a synchronous motor,
Fig . 2 presents a motor model,
Fig . 3 an analogue implementation of speed regulation in a synchronous motor,
Fig . 4 illustrates the principle and structure of synchro signal detection,
Fig . 5 illustrates a coordinate system transformation from a tπaxial to a biaxial system,
Fig 6 illustrates the rotation of a biaxial coordi¬ nate system, F Fiigg.. 7 7 presents a block diagram representing preven¬ tion of power stage saturation,
Fig . 8 presents an example implementation of preven¬ tion of power stage saturation,
Fig . 9 presents a regulator implementation, F Fiigg 1 100, 11, 12 visualise the equations in Fig 9,
Fig . 13 presents a simulator circuit,
Fig . 14 presents the responses to a 100A current step m id and lq regulators, Fig. 15 represents the speed consequent upon a torque step,
Fig. 16 presents the responses,
Fig. 17 represents the speed with a larger gam, and
Fig. 18 presents the signals in Fig. 1.
The block diagram in Fig. 1 represents a regulator for a permanent magnet synchronous motor. In principle, the problem of motor regulation has been solved if a work¬ able torque regulator exists. The regulator presented here is a torque regulator. The structure of the speed regulator controlling the torque regulator is not presented m detail. It may be e.g. an SCD as is conven¬ tionally used in fast KONE elevators. Preferably the
speed regulator can be implemented via software, using a PC or other processor as hardware.
In principle, the regulator structure consists of only two proportional controllers and two coordinate system transformers, so the device is quite simple.
The essential point about the regulator implementation in Fig. 1 is that both control loops are stable. This can be seen from the motor model in Fig. 2. The control loop contains only one integrator, so the regulator is always stable if implemented using proportional control In a proportional controller, the controlled variable always comprises a small static state error, but m this case it does not matter because the speed regulation loop is topmost . The gain needed m the controllers is obtained experimentally by finding the maximum allowed value that still allows stable operation of the device. When the gain is increased, the speed of the controller increases and the static error decreases. The speed is limited in the first place by the speed of the voltage regulator of tne inverter, and when it approaches this speed, regulation becomes unstable. In the power electronics output stage, the voltage regulator acts as a corrector of non-linearity of pulse width modulation. The voltage regulator contains a variable gam amplifier, to which a PWM-shaped voltage is fed back from the power stage. The regulator must contain a filtering delay to allow computation of the mean value of the modu- lated voltage. The reference value is typically an analogue signal obtained from the output of the torque regulator. The voltage regulator is generally the innermost control loop and therefore also the fastest regulator in the system.
Another important principle is zero adjustment of the longitudinal current (id) . This makes the crosscurrent
(iq) directly proportional to the motor torque because in this case the longitudinal flux generated by the permanent magnets remains constant and the torque is the product of longitudinal flux and transverse flux. Thus, the crosscurrent regulator (iq) is a torque regulator, as can be seen from Fig. 2.
The output quantity produced by the regulator electronics is a three-phase stator voltage reference. This is fed as a reference value into the voltage regulators of the inverter. Such signals are available e.g. in the KONE V3F inverter structure. However, it may be better to control the pulse width modulator directly, because this reduces the number of nested control loops by one. This means a higher regulation speed m the speed regulator.
The current is measured using current sensors. The current sensor may be incorporated in the inverter struc- ture . Fig. 18 presents the forms of the signals and quantities (v, ref , Torque, d, uq, iq, la, ua, q) prevailing m a given load situation.
Fig. 2 presents a model and the set of equations on which it is based. The letter s is a Laplace variable. The coordinate system d-q is thought of as rotating with the rotor of the motor. The d-axis, i.e. the longitudinal axis is parallel to the flux vector for the permanent magnets .
The regulation principle represented by Fig. 1 can be implemented using analogue techniques as illustrated by Fig. 3. The inverter feeding the motor is controlled by a torque regulator, in which alO is an iq regulator and a9 is an id regulator. The rest of the electronics mainly serves for coordinate system transformations.
In certain respects, the solution presented is still a solution m principle. The problem remains that the power stage is saturated before full speed is reached. Traditionally, this has been corrected by adding the third harmonic to the sinusoidal form of the phase volt¬ ages. In this case, as there is no three-phase oscillator, the correction has to be effected m some other way. This problem can be settled m a fairly simple manner e.g. using a suitable analogue solution.
For the rotation of the coordinate systems, a sensor is needed. A handy solution is a resolver mechanically connected to the rotor. The number of pole pairs is taken into consideration e.g. by using a suitable transmission ratio or by selecting a resolver with a number of pole pairs equalling the number of pole pairs in the motor. The coupling between the resolver and the motor must be so implemented that no slip occurs.
Fig. 4 illustrates the principle of phase-sensitive detection m the resolver.
Fig. 5 and 6 illustrate coordinate system transformations. The figures represent the transformation associ- ated with the measurement of current m the light of two sub-tasks. The reader is advised to figure out the other corresponding transformations along the same principles. According to the block diagram, there are two of these, and each transformation comprises two phases. Transfor- mation from a three-phase system into an equivalent two- phase system and then transformation of the two-phase coordinate system from a fixed system into a coordinate system rotating with the rotor and vice versa.
The three phases of a three-phase motor can be readily described using a vector diagram. A single vector describes all three phases at once. First, the motor must
be thought of as being reduced to a single pole pair version. The coordinate system is then placed on the end of the motor. The three axes are the directions of the three-phase winding. In Fig. 5, the instantaneous cur- rent values are obtained by projecting the current vector ι_ onto each one of the three axes . As the current vector rotates as a function of time, the instantaneous values of each phase vary. The same current vector can then be depicted in a two-axis coordinate system a, b as well. This coordinate system, too, s a fixed system m respect of the stator as seen from the end of the motor. The A-axis and the a-axis lie on top of each other and are coincident A two-axis coordinate system again is needed as an intermediate form to obtain a two-axis co- ordinate system d, q rotating with the rotor. This rotation is illustrated by Fig. 6. Further information on the basics of coordinate system transformations is to be found e.g. m the book "Emfuhrung m die Theorie gere- gelter Drehstromantriebe" by H.Buhler, ISBN 3.7643-0837- 0.
Power stage saturation is preferably prevented by shifting the zero point of the three-phase current when a phase voltage exceeds the maximum allowed value (=Uref). The shift is effected by amplifying the voltage by a coefficient k. The coefficient may be a fixed one or its magnitude may be determined separately as required in each case. This will not affect the main voltages. The block diagram in Fig. 7 represents the prevention of power stage saturation and Fig. 8 presents an example implementation for the prevention of power stage saturation.
The regulator can be implemented e.g. as follows. The circuit can also be built using multiplying DAC and PROM components. In this case, the matching of the control to the pole pairs of the motor can be effected via the
electronics, thus obviating the need for a gear. The inverter voltage is formed in the PROM circuits, m which the control has been matched to 19 pole pairs in the case of the example. The third harmonic can now also be easily added and the problem of inverter saturation is eliminated. The inverter also still comprises a voltage regulator, which has proved to be advantageous. Fig. 9 presents a solution of this type.
The equations presented m Fig. 9 are derived and/or illustrated Figures 10, 11 and 12.
The principle presented can be simulated as follows.
Fig. 13 presents a circuit which was used m the Simulab simulation program to verify the torque regulation described. The simulation input signal was a 100-ampere square wave used as a crosscurrent reference value. According to the theory presented, this should produce a torque reference value also in the form of a square wave, and actually this is one of the results obtained via simulation, as illustrated by Fig. 14.
Presented below are some of the simulation results . The curves were initially obtained with proportional gam values of 300 in the Iq regulator and 1000 in the id regulator.
Fig. 14 presents the responses to a 100A current step in the id and iq regulators . The upper curve represents the current id, which is to be adjusted to zero as far as possible. Below m the figure are the actual value and the reference value of the torque. Fig. 15 shows the speed response to a torque step. The curves obtained are quite convenient, and a good real device can be constructed using these values.
The next step in the simulation was to increase the gain to the value 1000. Fig. 16 presents the responses with increased gain.
The actual torque value and the reference torque value now coincide so that they cannot be distinguished from each other. It can be established that the torque faithfully follows the reference value, never exceeding it, for example. The operation is quite ideal.
The corresponding speed curve is shown in Fig. 17. The speed is a pure integral of the torque, as is obvious from the preceding curves .
It can be stated that by using the motor model and regulators described, a good torque regulator that is free of any propensity to vibration is obtained.
It is obvious to a person skilled in the art that the embodiments of the invention are not restricted to the examples presented above, but that they may be varied within the scope of the following claims.