Fuzzy-PID Methods for Controlling Evaporator Superheat

A new method is proposed to control the evaporator's superheat with an electronic-expansion-valve (EEV). The conventional PID control with invariable parameters can not receive sound performance because of the variation of evaporators' characteristic parameter under disturbances. To solve this problem, fuzzy algorithm is utilized to adjust parameters ofPID control on line, two practical methods are given and both are performed in experiments. It is found that fuzzy-PID greatly reduced the time to reach steady state, the error(±0.3"C) is much smaller than that of conventional PID, which is± 1·c.The hardware and software of fuzzy-PID are no more complex than those of conventional PID. By use of fuzzy-PID, the control performance is improved, but the cost does not increase. INTRODUCTION Control of the degree of evaporator superheat is an important aspect in refrigeration system control, because it affects efficiency and safety of system directly. The performance of a controller depends on the matching of controller's characteristic to the object. To get good control effect, it is necessary to be acquainted with object's characteristic. Generally, a first-order lag plus time delay model is used to describe an evaporator, it is found that the characteristic parameters in the model alter accordingly with the variation of operating condition. With theoretical analysis and practical experiments, the evaporator's response to variation of refrigerant flow rate and load is observed. Because of its simplicity, robustness and accuracy, PID control is often used for air conditioners and refrigeration machines. But the parameters of conventional PID will not change once they have been defined. At the same time, PID is a linear control method· and it can not take satisfactory effects on the highly nonlinear object. Fuzzy control does not need the accurate mathematical model of controlled object and it can also perform very well in multivariable, nonlinear and time-varying systems. But sometimes its accuracy does not meet requirement, it is only a coarse controller. In consideration of individual merits and drawbacks of them, in this paper, these two methods are combined, the fuzzy algorithm is implemented to adjust PID parameters on line in order to improve the applicability ofPID. EVAPORATOR DYNAMIC CHARACTERISTICS Some researchers think the response of superheat to variation of refrigerant flow in an evaporator is a K first-order lag process, G(s) = 1 + Ts , K and T change with evaporation temperature and refrigerant flow. Here, the evaporator is treated as a first-order lag plus time delay object, its transfer function can be expressed Eighth International Refrigeration Conference at Purdue University, West Lafayette, IN, USAJuly 25-28, 2000 337