Adaptive neuro-fuzzy control of systems with time delay

The authors present an adaptive fuzzy logic controller, which learns about the dynamic of the system under control from an online neural network (NN) identification algorithm. The identification is based on the estimation of parameters of a First-Order-Plus-Dead-Time (FOPDT) model. The outputs of the NN are three parameters: gain, apparent time delay and the dominant time constant. By combining this algorithm with a fuzzy logic controller with rotating rule-table, an adaptive controller is obtained which, with very little a priori knowledge, can compensate systems with time delay. The simplicity and feasibility of the scheme for time delay control provides a new approach for a variety of control applications. Simulation results are included to demonstrate the adaptive property of the proposed scheme