Design and Application of a Fuzzy Neural Network Forecast Controller

摘要

Superheated steam temperature in a drum boiler is a big-inertial and big-lagged process; moreover, its dynamic behavior varies greatly on different operating conditions. In this paper ,based on the characteristics analysis of the superheated steam temperature, a cascade feedback control strategy with load feed-forward is put forward. The control strategy consists of two loops. The master controller, as the outer loop in the control system, is a fuzzy neural network forecast controller, that is: the controller uses a neural network as the forecast controller, and the parameters and structure of the neural network are optimized simultaneously by improved genetic algorithm ; then it integrates fuzzy control with neural network to realize model predictive control of nonlinear system by neural network identification. A 2-order PID controller serves as the sub controller, which is applied to the inner loop of the control system to ensure a better stability and robustness. Simulation research shows that the cascade controller can markedly improve the control performance of the nonlinear and big-lagged system, and is easy to be applied too.