Discrete time adaptive neural network control for WME and compression refrigeration systems.

In this paper, we study the discrete-time control problem for the refrigeration system with unmodeled dynamics. This paper proposes discrete adaptive neural network controllers for two refrigeration systems, including the water membrane evaporator cooling and compression refrigeration systems. The influence of model uncertainty on system performance can be eliminated effectively by designing the neural network and the corresponding discrete-time adaptive updating strategy. Thus, the model parameters in the refrigeration system are not required in our control algorithms. Simulation results show that the proposed strategy can effectively adjust the refrigeration system temperatures. Compared with the traditional PID control strategy, the system response overshoot under our method can be reduced by 0.3%, and the system settling time is reduced by at least 94.3%. • Discrete-time adaptive neural controller is proposed for refrigeration systems. • Uncertainty problems are solved by using the adaptive scheme and neural networks. • Compared with the PID method, the settling time is reduced by at least 94.3%. [ABSTRACT FROM AUTHOR]