USE OF ANALOG COMPUTERS FOR TUNING CONTROLLERS IN INDUSTRIAL AUTOMATION

A. A. Gurchenok, V. S. Andyk, M. D. Kats

(Presented by the seminar of the Department of Automation of Thermal-Power Processes of Industrial Enterprises)

The broad introduction of automation into all branches of the national economy, now being observed, increases the requirements for the quality and reliability of the operation of all devices in automatic-control systems; this is determined to a considerable extent by the choice of the system structure and of its tuning parameters.

Because the entire control complex has become more complicated, the requirements for finding optimal controller settings have increased. Optimal controller parameters improve the qualitative and economic indicators of the operation of the main and auxiliary equipment and the reliability of their operation.

The study of the dynamics of automatic-control systems is a difficult problem. Determining the stability of low-order linear automatic-control systems is comparatively cumbersome. Studying the control quality of high-order systems is still more complicated, even when indirect calculation methods are used.

In studying nonlinear automatic-control systems, the difficulties increase many times over and, in a number of cases encountered in practice, may prove insurmountable with a purely theoretical calculation.

For calculating the optimal tuning parameters of industrial control systems with complex internal and external feedback connections, various methods have been developed [2, 3] that give satisfactory results, but that, as a rule, require cumbersome computations and ultimately additional correction when they are set on an actual controller.

There exists, however, another way to study automatic-control systems and to determine the optimal tuning parameters of controllers, one that, in combination with theory, ensures the solution of the above-mentioned problems. This way is modeling. Studies of automatic-control systems on analog computers (AVM) are possible by methods involving the assembly of the complete structural diagram of the automatic-control system or by solving the differential equation of the automatic-control system. However, it should be taken into account that, in such a formulation of the problem, we have an ideal model of the controller and, consequently, ideal tuning parameters, which, as practice shows, when set on an actual industrial controller do not always give satisfactory results.

Another, more acceptable way is the connection of a real

of an industrial controller with the model of the plant, set up on an analog computer. The structural diagram of the setup is shown in Fig. 1.

The output matching device of this scheme serves to convert the constant voltage of \(\pm 100\) V at the output of the analog machine into the input quantity of the industrial controller of the required physical nature (direct or alternating voltage, mechanical force, or air pressure) and the required range of variation. The input matching device converts the output quantity of the industrial controller into the input quantity of the plant model. Such a construction of the scheme makes it possible to set (or, by special engineering methods, to find) the optimal tuning parameters on a real industrial controller, while reproducing the controlled plant with a sufficient degree of accuracy on the modeling machine.

This method is beginning to find wide application for optimizing the parameters of controller tuning [1]. On the basis of this methodology, at TPI in the Department of ATPTP a training stand has been built for teaching students the method of tuning controllers and determining the tuning parameters of controllers of an industrial automatic control system. The training stand consists of an analog-type computing machine MN-7, a panel of controllers, and a recorder. The use of the MN-7 modeling machine makes it possible to simulate controlled plants described by differential equations up to and including the sixth order, as well as various typical nonlinearities, such as “backlash,” a “relay element,” etc. The sufficiently broad possibilities for plant modeling make it possible, in practice, to cover automatic control systems of drum boiler units and most automatic control systems of once-through boiler units.

Fig. 1. Structural diagram of coupling the plant model, set up on an analog computer, with a real industrial controller

The stand itself is a panel on which are mounted an industrial serial controller of the RPIB type, a DL-P differentiator, a model of the actuator MIM, a serial correcting controller for multiloop control of the KPI type, a board of the stand’s patch field for implementing the necessary variants of the circuit, a BU-1/6 control unit, indicators of the position of the regulating element and of the output of the correcting instrument. Of the elements of the stand, the actuator model is of particular interest. The transition to the actuator model is due to the bulkiness of serial specimens for use for modeling purposes.

The circuit of the actuator model is shown in Fig. 2.

The actuator model is a reversing actuating motor of the RD-09 type, powered by a reduced stabilized voltage from transformer TR-1. Reversal of the motor is carried out by means of the contacts of the repeater relay of the controller.

of the RPIK type, or a trigger with power amplification at the output amplifier of an RPIB-type controller. The output voltage of the actuator model is determined by the imbalance of the resistance bridge \((R_5, R_6)\). The resistance bridge is supplied with rectified, well-filtered, stabilized voltage. The time constant of the actuator model is 30 seconds, which is an average value for real actuators. The output signal of the automatic control system is recorded by a self-recording potentiometer, which makes it possible to obtain a clear record of transient processes.

The training stand was used to determine the tuning parameters of automatic control systems for the boiler house of the Tomsk DOK. As an example of the calculation, an automatic control system for the water level in the drum of the boiler units was selected. The controlled object is a series connection of a capacitive and an integrating element and is described by a second-order differential equation. The data on the controlled object were represented by experimentally obtained time characteristics. After their statistical processing and approximation, it became possible to give a mathematical interpretation of the controlled object with its subsequent modeling on the MN-7 analog computer. The tuning parameters of the controller were found by the engineering method of stepwise search [1], which is usually used in the adjustment and correction of industrial automatic control systems.

Fig. 2. Electrical circuit of the actuator model (MIM).

At the same time, for verification and comparison of the results obtained, the tuning parameters of the controller were calculated by the graph-analytical method [2].

The transient processes for the step-search method and the graphoanalytical method were recorded by a self-recording instrument and are presented in Figs. 3 and 4.

Analysis of the transient processes shows that the controller tuning parameters obtained by the step-search method provide better quality of the transient process than the graphoanalytical method. The dynamic error and the settling time of the transient process are reduced, which is very important for real systems. Setting these parameters on the real system proved their full correspondence to normal operation of the automatic control system.

Fig. 3. Transient processes in the automatic control system through the control-action channel: 1 — the optimal tuning parameters were found by the graphoanalytical calculation method; 2 — the optimal tuning parameters were found on the trainer stand by the step-search method

Fig. 4. Transient processes in the automatic control system through the disturbance-action channel. The designations are analogous to Fig. 3

When determining the tuning parameters on the stand, the great visual clarity of the adjustment operations being performed comes to the fore. With the slightest change in the controller tuning parameters, the system model immediately responds with the corresponding transient process, the quality

which can immediately be determined by known methods [3], whereas in analytical calculation the quality of transient processes is determined by further graphical constructions and analytical calculations. Such simplicity and speed in obtaining transient processes make it possible to use the modeling stand for teaching students how to tune a controller of practically any type.

REFERENCES

1. Instructions for tuning and testing electronic controllers of block installations. ORGRES, Novosibirsk, 1970.

2. V. Ya. Rotach. Calculation of the tuning of industrial control systems. Moscow, GEI, 1961.

3. E. P. Stefani. Fundamentals of calculating the tuning of a controller for thermal-power processes. Moscow, “Energiya,” 1972.