Research Of Billet Temperature Modeling And Optimization For Reheating Furnace Based On D-FNN

Reheating furnace is an important thermal technology equipment in the process of billet rolling. It can assure the rolling quality and reduce energy consuming in heating furnace if the billet's temperature is control reasonably. What's more, the shortage of energy sources is seriours nowadays. Therefore, it is significant to establish a valid model of the furnace, calculateing the outlet strip temperature.In this paper, based on walking beam regenerative reheating furnace system, there are a brief overview of regenerative reheating furnace and intelligent algorithm. In this paper D-FNN(Dynamic Fuzzy Neural Networks) model proposed is used to the billet temperature forecasting. In this paper, accoding to reheating furnace's production data, how to apply and optimize D-FNN to build predictiong model of billet temperature had been studied. Firstly, the model of billet's temperature prediction with D-FNN has proposed. And the D-FNN for modelling the billet temperature prediction has been verified through simulation results. Secondly, the main parameters of D-FNN has been optimized. This paper used the EKF(Extended Kalman Filter) to optimize the parameters of the premise, and used the LLS(Linear Least Squares) optimize the parameter of the results. This paper observed the experimental results of prediction model base on EKF and LLS optimizing parameters by simulation. For further improving the model's accuracy and speed of recognition, through comparing simulation result of billet's temperature prediction model found that the parameters of results adjustment method should be improved. This paper tried to use PSO(Particle Swarm Optimization) instead of LLS to optimize the parameter of the results. Next, the results of simulation shows that the convergence rate of PSO algorithm slow down in the later period of optimization iterative, and PSO easy fall into the local limit. For improving the PSO algorithm, the artificial immune clonal algorithm is used, and put forward an optimization method of PSO with IMMCSA(Immune Memory and Multi Clonal Selection Algorithm). This paper use IMMCS A-PSO adjust the parameters of result, and use improved net to model the billet temperature forecasting.Simulation results showed that the experimental results of prediction model based on EKF and IMMCSA-PSO optimizing parameters by simulation were good, and the error had been reduced to the extent that expected. Based on the predicted billet's temperature, the operator could control and treatment in a timely manner to ensure the quality of billet. Therefore, the improved optimization method D-FNN model not only had theoretical significance, but also had practical value.