Rough set-neural networks based neural PID control of generator excitation system

Rough set is a powerful mathematics tool with merits of intelligent data analysis and rule extraction, and Radial Basis Function (RBF) neural network has the ability to approach any nonlinear function precisely. An adaptive neural PID control strategy based on integration of rough set theory with RBF neural networks is presented for synchronous generator excitation system. The reduced decision rule set, which is acquired through rough set intelligent data analysis, is used to configure RBF neural networks by Orthogonal Least Squares (OLS) algorithm. Then the parameters of neural PID controller are tuning according to rough set-RBF networks model identification on line. The controller designed here can map the nonlinear characteristic of excitation system, and the dynamic response of generator. The simulation results demonstrate that the proposed method is much more effective than conventional PID control for improving dynamic performance and stability under small and large disturbances.