An adaptive fault tolerant PI controller for an information-poor, nonlinear system

Abstract

In this paper, an adaptive PI controller is proposed to achieve fault tolerant control. Information-poor systems are normally nonlinear, and difficult to be described via physical models. Hence fuzzy relational model is used to describe their behaviours with only input/output information. An aircooling subsystem in HVAC (Heating, Ventilating and AirConditioning) system is used as a case study. The neurofuzzy based adaptive PI controller is designed to accommodate the faults in the system with fault tolerance achievement. Simulation experiments have been performed to prove the efficacy of the proposed algorithm.