Nonlinear Correction for Improving the Quality of Transient Processes in the MUS-D System of a Deeply Controlled Electric Drive

A. P. Ineshin, V. A. Sevastyanov

(Recommended by the scientific seminar of the Electromechanical Faculty)

Regulated DC drives with MUS of the PMU-M series, developed by ENIMS, have found wide application for the speed-control range of the actuator motor \(D = 10 : 1\). At present, prototypes of PPMU drives have also been developed and produced for the control range \(100 : 1\) [1], with the introduction of speed feedback from a DC tachogenerator through an intermediate semiconductor amplifier (PPU).

Further expansion of the control range, while simultaneously improving the dynamic and static qualities of the drive, can be achieved by using a PPU of a more advanced type and the corresponding circuit designs of the automatic-control system.

It should be noted that improvement of the static accuracy of an automatic-control system by increasing the gain coefficient of the controller is limited by the dynamic properties of the controlled object. One possibility for increasing the overall transfer coefficient of the controller without loss of stability and without a substantial deterioration of the other dynamic properties of the system is the use of correcting devices, in particular, the introduction of derivatives into the control law.

However, for automatic-control systems containing correcting links with constant parameters, there exist limiting quality indices, which decrease considerably when the characteristics of the system elements are nonlinear and when a limitation is introduced on the permissible acceleration of the controlled variable [2]. These difficulties can be overcome by creating automatic-control systems with a variable structure, in which, in order to improve the dynamics, the correcting circuits are switched depending on the state of the system. Some possible laws for changing the parameters of correction circuits and the corresponding control systems are considered in [3].

The Department of Electric Drive of the Ulyanovsk Polytechnic Institute, in collaboration with the Ulyanovsk Heavy Machine Tool Plant, has created and prepared for implementation a version of a nonreversible MUS-D drive with PPU for the speed-control range \(D = 200 : 1\); in a number of dynamic indices, it is not inferior to an analogous electric drive

according to the EMU-D system with an IPU. A simplified structural diagram of this electric drive is shown in Fig. 1, where ID is the actuating DC electric motor, MUS is a static current converter implemented on the basis of a standard power magnetic amplifier with self-saturation; IPU is an intermediate semiconductor amplifier of class D. Increasing the speed-control range to \(200:1\) is accomplished by introducing rigid negative speed feedback from the DC tachogenerator TG through the IPU. To stabilize the drive, flexible negative feedback is used,

Fig. 1

embracing the links with the greatest gain and the greatest inertia: the IPU and the MUS. For this purpose, in order to obtain the stabilization signal of the controlled quantity, a separate control winding of the MUS is used, whose signal is proportional to the derivative of the resultant magnetic flux of the cores [4]. In order not to change the static gain coefficient of the system, the correcting feedback must necessarily have a typical element of the form

\[ \frac{K_k pT_k}{1+pT_k}, \]

which at very low frequencies (steady-state mode) as it were eliminates the action of the feedback, while at frequencies where

\[ \frac{pT_k}{1+pT_k} \approx 1, \]

it transforms it into rigid negative feedback with correction transfer coefficient \(K_k\).

The operation of the automatic control system under consideration (IPU—MUS—D) is distinguished by the presence of two possible modes, caused by saturation in the direct channel of the system [5]. In view of the large gain coefficient, the inertia-free IPU becomes saturated already at small amplitudes of the error signal, thereby weakening the action of the correcting feedback. Outside the zone of linearity, under large external disturbances, the correcting feedback becomes so weak that the correction loop turns out to be as if open: the automatic control system, open along the correction circuit, operates in a mode-

... of optimal control; in this case, maximum use is made of the capabilities of the actuating motor and, consequently, maximum speed of response is obtained for the given automatic-control system. The transition from the optimal-control system to a linear-control system with an effectively operating correction circuit is necessary in the small-signal zone because of the danger of self-oscillations near the equilibrium position. For the automatic-control system in Fig. 1, saturation is a typical mode; it operates according to the principle of two-zone control and, in the region of large external actions, does not require a special switching device for transferring the system from one mode to another, since a natural transition is provided from the saturation region to the linear zone and back [5].

Large external actions that cause a change in the operating mode of this system include practically the entire range of control signals (owing to the stepwise execution of the speed setter) and part of the disturbing actions when the drive operates in the upper and middle part of the speed-control range. Therefore, the transient processes of the automatic-control system in the indicated regions of action proceed quite satisfactorily.

In the lower part of the range, operation of the automatic-control system takes place predominantly in the linearity zone, which causes an unsatisfactory course of the transient process. This is confirmed by the oscillogram of the transient process in the drive shown in Fig. 2, operating at a speed \(n = 30\) rpm (\(D = 100 : 1\)) under normal load and its subsequent removal (dropping). The principal indicators of the transient process, obtained from the oscillogram, are summarized in Table No. 1a. It is characterized by oscillatory behavior, large values of overshoot, and is prolonged in time. This is explained by the fact that, first, when operating at low rotational speeds of the AM, the effectiveness of the action of the main rigidity circuit—

Fig. 2.

of negative feedback is insignificant, owing to the small magnitude of the reference voltage and the emf of the tachogenerator; secondly, there is no dynamic-braking loop, which makes it impossible to dissipate the stored energy of the dc drive when overshoot occurs; and, thirdly, the flexible correcting feedback acts strongly. This last circumstance has a decisive effect on the quality of the transient process in the drive when the load is shed. In this case the current of the armature circuit and the motor torque proportional to it, because of the action of the correction circuit, do not decrease immediately but smoothly, which leads to the appearance on the motor shaft of a dynamic torque causing acceleration of the motor-shaft rotation. The resulting initial overshoot is large, which causes the subsequent oscillatory course of the transient process.

A satisfactory transient process when the load is shed in the system of Figs. 1 and 3 is ensured by introducing a nonlinear discrete-correction link (NDK), which makes it possible to obtain the previously mentioned limiting control mode of the automatic-control system also in the lower portion of the speed-control range of the dc drive. This is achieved by artificially disconnecting the parallel-correction circuit for the duration of the transient process when the load is shed. The principal elements of the NDK link (Fig. 3) are:

K — a switching element which disconnects the correction signal by shunting its circuit;

P — a threshold element which performs selective control of the switching element K only for the duration of the transient process of load shedding in the dc drive.

Fig. 3.

The schematic diagram of the NDK link is built entirely on semiconductor devices (Fig. 3). As the switching element “K,” a p—n—p semiconductor triode is used, normally closed in the absence of a control signal. A semiconductor diode is included in the collector circuit of the triode in order to ensure the operating polarity of the triode collector voltage. The threshold element “P”

is a stabilitron, whose connection point determines the actuation setting of the NDC element. To control the NDC element, a signal from the correcting circuit is used, of a definite magnitude and polarity corresponding to the transient process in the electric drive when the load is shed.

With negative polarity of the signal relative to the common lead, which corresponds to the transient process of MUS locking, the stabilitron breaks down, opening the semiconductor triode through the base circuit. The latter passes from the cutoff mode into saturation, shunts the correction signal, disconnecting it from the input circuit of the CPG, and thereby ensuring the limiting speed of response of the ACS.

In the transient process of MUS unlocking, which corresponds to positive polarity of the signal in the correction circuit, the NDC element does not operate, since the valve in the collector circuit of the triode and the stabilitron are closed. Breakdown of the stabilitron in the reverse direction is possible only during the initial connection of the MUS to a three-phase alternating-current network. When the power circuit is switched in this case, an overvoltage pulse of positive polarity arises in the correction circuit and breaks down the stabilitron. In order to eliminate the reverse voltage at the locked base-emitter junction of the semiconductor triode, the latter is shunted by a valve. In other cases of actuation of the NDC element (overspeed regulation during starts, transfer of the ED from high speed to low speed, or load-shedding mode in the middle and upper parts of the range), its influence is insignificant, since in these cases the ACS automatically passes into the mode of limiting speed of response.

The beneficial action of the NDC element when load is shed at a low rotational speed of the ED is confirmed by an oscillogram of the transient process recorded in an actual drive under the same conditions (Fig. 4). The principal indices of the transient process obtained from the oscillogram are given in Table No. 16. Now, owing to the action of the NDC, which disconnects the correction circuit when the load is shed, the armature current and mo—

Fig. 4.

the motor torque decreases considerably faster than without the NDC element (Fig. 2). This accounts for the small magnitude of the initial overshoot of the AM speed, the subsequent rapid damping, and the shorter duration of the transient process.

It should be noted that the introduction of nonlinear correction of this type does not worsen the indices of other types of transient processes. The presence of valves in the converter power circuit and the impossibility of electrical braking of the motor do not impose any limitations on the magnitude of the practically attainable converter-locking speed. However, if several nonlinear correcting devices are used, for example to improve simultaneously the quality of transient processes when load is applied, negative consequences may occur. In this case the transient processes and the operation of the correcting devices will prove to be mutually coupled and, under unfavorable circumstances, deterioration of quality, loss of stability, and transition to a regime of hard oscillations are possible in the system.

The range of application of the method of introducing NDC according to the circuit in Fig. 3 can be considerably expanded. Without substantial modifications it is applicable, for example, in deeply controlled electric drives with a nonreversible ionic or semiconductor converter.

Table No. 1a

Table No. 1b

REFERENCES

1. A. Ya. Petrovsky, Ya. B. Rozman. Regulated Electric Drive with Magnetic Amplifiers. “Energiya,” 1964.

2. Fundamentals of Automatic Control. Edited by V. V. Solodovnikov. Mashgiz, 1954.

3. V. A. Taran. Review: “The Application of Nonlinear Correction and Variable Structure for Improving the Dynamic Properties of an Automatic-Control System.” Avtomatika i telemekhanika, vol. XXV, No. 1, 1964.

4. M. A. Rozenblat. Magnetic Amplifiers, “Sovetskoye radio,” 1960.

5. Analysis and Design of Nonlinear Automatic-Control Systems. J. T. Tou and M. Pestel. “Energiya,” 1964.