Predictive Control Of BTP Based On GA-PSO-BP Neural Network With LM-GA Algorithm

Sintering process is not only a basic link of iron and steel metallurgy,but also a very complex nonlinear dynamic time-varying process.The burn-through point(BTP)refers to the position where the sintering mixture is completely burned through,which can be represented by the air box of the sintering machine.As an important parameter of sintering state,BTP can directly show the state of sintering process,and has an important influence on the quality and output of sinter.It is difficult to predict and control the BTP accurately by using the mechanism modeling and traditional control methods.Aiming at the problem that it is difficult to predict and control the end point of sintering,this paper studies the prediction model of the BTP and the prediction controller of the BTP.The specific research is as follows:Firstly,in this paper,based on the analysis of the characteristics of sintering process and the factors affecting the BTP,five major factors are determined: the operating speed of the sintering trolley;temperature of(No.17)air box at BRP;the ignition temperature;the thickness of the feeding layer and the moisture content of the second mixture.Soft sensing model of BTP was established by using the temperature of exhaust gas bellows.Because BP neural network has a strong approximation ability to the nonlinear system,BP neural network is selected to identify the BTP model.In order to improve the performance of BP,GA-PSO hybrid algorithm is used to optimize the weight and threshold of neural network,and a GA-PSO-BP neural network BTP prediction model is established.Simulation results show that the prediction model has high prediction accuracy.Then the L-M algorithm is more dependent on the initial value,this paper uses GA algorithm to improve the L-M algorithm,then the advantages of GA algorithm's strong global convergence ability to provide high-quality initial value for L-M algorithm,so as to improve the calculation speed of L-M algorithm,also overcome the lack of L-M algorithm's too dependence on the initial value.The simulation test shows that LM-GA algorithm's optimization performance is better.Finally,the design of the predictive controller for the BTP is studied.GA-PSO-BP is used to establish the prediction model of BTP.LM-GA algorithm is used in the rolling optimization part of BP neural network prediction controller to obtain the optimal initial control value trolley speed.The controller controls the BTP through trolley speed.This optimization method improves the performance of the prediction controller of BTP.Analysis of simulation results,it is concluded that the performance of the predictive controller designed in this paper is better than that of the conventional predictive controller and PID controller,which has faster response speed,shorter regulation time,stronger anti-interference ability and robustness.