Research On The Prediction Model Of Steel Temperature For Reheating Furnace

The reheating furnace is an important thermal equipment in the steel rolling process,and its main function is to heat the billet to the temperature required by steel rolling.Reasonable billet temperature can not only improve the quality of billet,but also reduce energy consumption.So it is necessary to establish a precise temperature prediction model for billet.At present,the model of steel temperature prediction mainly consists of mechanism model and neural network model.The type of neural network has some influence on the prediction accuracy of the prediction model of steel temperature,and the comparative study is less at present.In this thesis,on the basis of studying the modeling method based on total absorption rate for prediction of billet temperature,the shortcomings of the model are analyzed.In view of the many parameters of the mechanism model,complex calculation and difficult selection of parameters,the modeling scheme based on neural network is determined.First,the temperature prediction model of BP neural network is established and simulated on the basis of heated a section,two sections of heating,and the heat equalizing section of reheating furnace.In order to solve the problem of insufficient precision of simulation results,this thesis uses particle swarm optimization and swarm optimization to optimize the weights and thresholds of the BP neural network respectively.In this thesis,the corresponding algorithm optimization BP neural network flow is designed.The experimental results show that the two kinds of optimization can improve the prediction accuracy of the model.In order to solve the shortcomings of BP neural network modeling,such as easy to fall into local optimum and over fitting,a prediction method of steel temperature based on extreme learning machine(ELM)is put forward.The ELM neural network is introduced into the prediction model of steel temperature,and the algorithm flow is designed.The number of the best hidden layer nodes is determined by the golden section method.In order to further improve the prediction accuracy,particle swarm optimization and artificial bee colony algorithm are used to optimize the ELM method respectively,and the PSO-ELM prediction model and ABC-ELM prediction model are established,and the simulation research is carried out.Through the comparison and analysis of various models of steel temperature prediction and the results of simulation experiments,the following conclusions are obtained:The common BP neural network has relatively poor effect,Particle swarm optimization BP neural network model and artificial bee colony algorithm optimization BP neural network model in prediction accuracy is better than the common BP neural network model,and the bee colony algorithm optimization neural network model has the best effect.For the steel temperature prediction model based on the limit learning machine method,the prediction accuracy is better than the BP neural network model,and at the same time the speed is faster.The ABC-ELM prediction model optimized and built by artificial bee colony algorithm has more accurate prediction results and stronger generalization ability.It is the most ideal prediction model for steel temperature in several prediction models.