GA Optimized Fuzzy PID Control with Modified Smith Predictor for HVAC Terminal Fan System

Abstract

The terminal fan is an important part of the Heating Ventilation Air Conditioning (HVAC) system, which undertakes the task of transferring heat between the circulating fluid and the room air. Unfortunately, the large time-delay is inevitable because of the heat transfer process. Recent years have seen the successful applications of various artificial intelligent algorithms in solving optimization and control problems for complex systems. Therefore, it is of significant meaning to design a machine learning algorithms based intelligent control method to balance heat transfer efficiency, fan speed and energy consumption. In this paper, a second-order terminal fan system model with pure delay is constructed. Ensuingly, PID control is applied to control the fan's speed, and fuzzy inference is used to adjust PID parameters. The fuzzy rules are optimized by genetic algorithm with the objective function designed according to performance indexes. The temperature difference control strategy guarantees heat transfer efficiency and energy saving of terminal fans. A modified Smith predictor is proposed to solve the time-delay mismatch problem further. The terminal fan system control process is simulated with different time delays, to confirm the robustness of the control algorithm.