Research of room temperature for inverter air conditioner based on Fuzzy Neural Network Control

Abstract

Keyword: Fuzzy control; Neural networks; Inverter air conditioner; Energy saving. Abstract. The indoor temperature control process in air-conditioned room is a multiple-input multiple-output (MIMO), nonlinear and time-delay system. Whereas the traditional off-on control methods, traditional PID control and conventional fuzzy control, exist some advantages and disadvantages, so this paper has provided the fuzzy neural network control modality to improve the current control approaches. Analysis on the simulation of the model as mentioned above is established, and simulation results show that the fuzzy neural network control has a series of positive qualities such as rapidity, good stability and strong anti-interference, so as to achieve a good temperature regulation in inverter air conditioner rooms. 1. FOREWORD With the development of society, the component of inverter air conditioner market has claimed an ever-growing in the market. Therefore, the research of the inverter air conditioner is becoming more and more significant. The new inverter air conditioner standard was promulgated on June 1, 2013, the threshold of inverter air conditioner energy efficiency limit has increased to 3.9, the energy consumption lower of inverter air conditioning become a focus point for each manufacturer and universities. At present, the inverter air conditioner control mainly based on PID control. Under the accurate mathematical model, to some extent, with its simple structure, good stability, reliable operation, easy to adjust, PID can make accurate control come true. But, the model is difficult to be constructed, with large lag, nonlinear, time-varying complex characteristics of air conditioning refrigeration system control object, and it makes it difficult to achieve ideal control effect for traditional PID. Fuzzy control system has advantages of quicker response, smaller overshoot and less sensitive to parameter variations. It relies on the language rules by expert control experience, so it shows good robustness and more effective for the nonlinear and complex controlled member. The neural network has the characteristics of self-organizing, adaptive, self-learning, it solves the problems depended on experts of the conventional fuzzy air conditioning temperature control mutation, membership function and fuzzy control rules. If combining the fuzzy control with neural network as the air conditioning temperature control, not only solved the puzzle of controlled member by the difficulty to precisely define its mathematical model, but also solved the puzzle of fuzzy control moderate depended on expert experience too much. And it makes the control system of autonomous learning ability realize intelligent control