Research On Neural Network Inverse Control Of Direct-fired System With Duplex Inlet & Outlet Ball Mill

The direct-fired system with double-inlet and double-outlet ball mill(DIDO-BM)has become more and more popular in domestic coal-fired power plants.The direct-fired system requires that the DIDO-BM output pulverized coal is equal to the boiler coal consumption.Because the direct-fired system with the DIDO-BM is a multivariable,large time-delay,nonlinear system,the conventional control methods are often difficult to meet the requirements of control quality.Therefore,it is important to design a suitable control system for the object.Based on the above questions,this work has done the following research:1.The mechanism model of the direct-fired system with the DIDO-BM has been established in this work firstly.By the prediction error minimization method,according to the historical operating data of a direct-fired system with the DIDO-BM,the unknown parameters in the model have been identified.By comparing the established model output with the actual output,the correctness of the established model is verified.The dynamic performance analysis of the model further proves the rationality of it.2.An inverse control scheme based on RBF neural network has been designed for the controlled object in this work.Because of its steady-state error and poor robustness,combining the idea of model predictive control(MPC),an inverse control scheme of RBF neural network based on MPC is designed,which makes the improved inverse control strategy of neural network better.The results of the simulation research show that the proposed control system performs better than the comparison methods.3.To avoid the the blindness of the proposed RBF neural network inverse controller's parameter settings,a fuzzy self-tuning particle swarm optimization(FST-PSO)is designed to solve the problem that the standard particle swarm optimization(PSO)solves the questions slowly and is easy to fall into local optimum.The simulation research on the controlled object shows that the controller using FST-PSO algorithm further improves its performance.