Bed Temperature Control Based On Neural Network Identification Of Circulating Fluidized Bed Boiler

Circulating Fluidized Bed Boiler( abbr. CFBB or CFB Boiler) is becoming the main coal combustion technology all over the world for its attractive advantages of high combustion efficiency, efficient sulfur removal, low NO_x emission,fuel flexibility and good load-following capability. High demand of automatic control of CFBB is proposed with its fast commercial development.But, CFBB is a new type boiler with immature technology till now. Some complex characteristics such as non-linearity, uncertainty, time-variety, undue lagging and multi-variable coupling. Study of combustion mechanism and modeling are not mature and model applied in real-time control has not appeared. So the low control level of the bed temperature, as an important parameter of circulating fluidized bed boiler, is keeping.In this thesis, through the analysis on mass balance and energy balance in the dense phase and sparse phase, the bed temperature cybernetic model is established in this thesis. Two types neural networks are designed: the single neron networks contoller and the RBF-Radial Basis Function controller. The neural network control system is simulated on the bed temperature of the circulating fluidized bed boiler. And then on the single neron PID reference model self-adaptive control system based on RBF neual network is designed. The simulation shows that reference model sel-adaptive control system have better performance on the identification and the trace of plant model.