Steam Temperature Control System In The Boiler Based On The Neural Network

Coal-fired power is the main power source in China. The boiler is one of thethree major equipments in thermal power plant in China. The stability andreliability of the boiler is very important to power generation efficiency. Themain steam temperature control system is the emphasis and difficulty in theboiler control system. The super-heater, as the controlled object, is workingunder high temperature and pressure. Because the continued stability of boilermain steam temperature is very important for keeping the unit operating safelyand economically, it has a higher requirements to the main steam temperaturecontrol system.This thesis puts direct-fired pulverized coal boiler in thermal power plant inChina as the research object, which is designed on main steam temperaturecontrol system for power plant boiler based on Radial Basis Function (RBF)neural network. Firstly, this paper introduces the basic principle and trainingmethods of RBF neural network, and describes the application of radial basisfunction neural network in the control system. Secondly, this paper analyzes thedynamic characteristics of the main steam temperature control system. Becausethe mechanism model parameters vary with the different working conditions, itproposes model identification of steam temperature control system based on RBFneural network and identifies the numbers of the hidden layer nodes, the centervalues and width values of radial basis function, the weights from the hiddenlayer to the output layer with the improved hybrid learning algorithm. Then ituses the MATLAB software to simulate the model identification system andanalyzes the generalization of the identification results. Thirdly, in order toovercome the disadvantages of the adaptive effects being poor by theconventional PID controller, this paper designs the RBF neural network PIDcontroller and gives the output curve simulation and comparison on step response. And it has been proved that this control strategy provide a promising prospect.