MPC, Controller tuning, Compressor test, Test rig, MIMO
This paper presents a practical methodology for tuning the parameters of a model predictive control technique for controlling the suction and discharge pressures of refrigeration compressor in test rigs. Typically, in this type of rig, the compressor under test is subjected to similar conditions as the ones found in refrigeration systems, such as refrigerators and freezers. Even though in industrial practice it is common to find proportional-integral (PI) controllers in such rigs, they are multivariable processes with partial coupling between the variables of interest. Model-based predictive control (MPC) is a control technique which uses an explicit process model to predict the future behavior of the system over a horizon, and then calculates a sequence of control actions so that the future outputs of the process track future references. Thus, since MPC is inherently a multivariable control technique, it can be used, for example, to mitigate the coupling between suction and discharge pressures, thus improving the performance of the control of the compressor operating condition and, therefore, the overall test performance. The study presented in this paper is based on a specific test rig used in industry, but the ideas are presented in a general way so that they can be used as general guidelines for tuning MPC for refrigeration compressor test rigs. The rig considered for this paper has two outputs, which are the pressures at the inlet and outlet of the compressor under test, and two manipulated variables, which are two valve openings. The paper begins by showing how to identify the dynamic models that describe the behavior of the compressor pressures and also how to use them to define an expression that relates the static gains of the models identified with the parameters of the predictive controller, with respect to closed loop-performance specifications of the test. In this study, the model predictive control technique known as generalized predictive control was used and, in addition to the tuning methodology, an analysis of the effects of the controller parameters on the closed-loop results is presented. Finally, the performance of the predictive controller tuned according to the proposed methodology is compared to the results obtained by two independent PI controllers, showing the improvement of the responses for both reference tracking and disturbance rejection. The obtained results are promising and show that the proposed methodology can be used as a starting point for the tuning of predictive controllers applied in test rigs. In addition, it is shown that the use of advanced control techniques, such as model-based predictive control, can contribute to increasing the productivity and operational efficiency of compressor tests.