Evolving Fuzzy Model-based Adaptive Control

The paper describes an evolving fuzzy model-based adaptive controller (eMAC) that is suitable for use in non-linear, uncertain systems. Two fuzzy models are used to predict the future behaviour of the plant; one is an evolving T-S fuzzy model that is learnt online from normal operating data; the other is a fixed T-S fuzzy model that is identified off-line from data obtained from a generic linear model of the plant to be controlled. The controller is applied to a simple non-linear dynamic system that has a significant time delay and simulation results are presented which demonstrate that the evolving fuzzy model-based adaptive controller does improve the performance of the control system. The controller is now to be tested experimentally on the air temperature control loop of a cooling coil in a real air-handling unit.