Research Of Control Policy On Main Steam Temperature Of Thermal Power Plant Based On Fuzzy Neural Network

The main steam temperature exported from the last stage superheater in modernplants (hereinafter referred to as the main steam temperature) is a very importantparameter for both monitoring and controlling the operation of the thermal powergenerating unit. It will directly affect the safety and economic behavior of the unit if themain steam temperature is too high or too low. Therefore, aiming at the thermalengineering characteristics in the thermal power plant so as to take reasonable andeffective control for the desirable main steam temperature and ensure higher fuelefficiency is a headache and hot spot for the control system at present.As the main steam temperature system has the characteristics of nonlinearity, timevariation, large delay and great inertia, it is not ideal to use the traditional controlmethod which must establish an accurate mathematical model for the system. Whileintelligent control methods can solve this problem effectively and take efficient globalcontrol for such complex system since they not only can analyse and synthesize thecontrol system from the function and the point of view of the whole optimization butalso do not rely on the mathematical model of the controlled object. So using theintelligent control methods for the main steam temperature control is an inevitable trendof development. At present, the main methods applied to the main steam temperaturesystem include fuzzy control, neural network control, fuzzy neural network control andgenetic algorithm control, etc.According to the main problems existing in the main steam temperature controlsystem, this paper proposed a composite intelligent control strategy based on the fuzzyneural network control technology. It keeps the original cascade PID control system andbrings in some intelligent control methods such as fuzzy neural network and RBF neuralnetwork and so on. This composite intelligent control strategy uses the experience ofhuman knowledge and can increase the intellectuality of the system so as to improve thecontrol quality of the whole system.In order to maintain the advantages of the cascade control system, the vice loop ofthe designed control system adopt proportion regulator (namely P regulator) realizingthe inhibitory action for the various internal disturbances. Meanwhile, the neural network feed-forward controller has been introduced in the vice loops which make upfor the big delayed shortcoming of the feedback control and solve this tough problem.Fuzzy neural network adaptive PID controller as a substitute for traditional PIDcontroller has been adopted in the main loop of the conventional cascade control systemso as to realize the inhibitory action for the various external disturbances. A reversesignal has been provided for the fuzzy RBF neural network by the use of the neuralnetwork identification controller to realize the function of online learning, self-learningand self-adaptability of the fuzzy neural network. This composite intelligent controlstrategy not only can adjust the control parameters of the system rationally also ensurethe ideal control effect in various conditions for the main steam temperature system.