Analysis and experimental study of MIMO control in refrigeration system

For various operating conditions and loads of refrigeration systems, converters are used to change the compressor’s speed, and electronic expansion valves are adopted to regulate refrigerant flow rates. The refrigeration system, as a process under control, is a multi-input multi-output (MIMO) and nonlinear complex system. Based on the analysis of fuzzy and neural networks control, a self organized fuzzy neural network controller with the capacity of construction and parameter learning is proposed according to the simplified fuzzy control algorithm and the similarity between structure and function of a counter propagation network (CPN). This controller is easily structured with the feature of fuzzy control, and it also possesses the learning ability of neutral networks. The air cooled refrigeration experimental results show that the present controller can modulate the evaporation pressure and the superheating.