Hot Strip Thickness Prediction Based On General Regression Neural Network

Steel is the most basic material supply for all industrial industries.The steel industry is great significance to a country's economic development and national defense capabilities.As the steel industry's largest proportion of products,it is also the product with the most profit.The industry has a wide range of applications.Rolling force is the most important factor that affects the quality of the hot strip,which not only affects the thickness of the product,but also affects the quality of the product.The traditional rolling force model is a mathematical model obtained by prior experience and physical and mathematical analysis.Due to various reasons of the model itself,it cannot achieve the very high precision value required.With the progress of industrial intelligence,integrating machine learning into steel rolling has become an inevitable trend,but which tool to choose to build a new model has also become a factor that must be considered.This article mainly uses the Generalized Regression Neural Network(GRNN)to establish a new rolling force prediction model.This algorithm is better than other algorithms for the prediction of the non-linear function of the rolling force,and is aimed at the parameters in GRNN.The Drosophila optimization algorithm is used for processing,so that the network can quickly find the optimal solution.The data all come from the actual offline data value of 1700 mm from the actual steel mill.The information corresponding to each steel is proposed using VS programming,and then the data is put into the SQL database,and then the principal component analysis method is used to analyze the data Dimensionality reduction processing,put the processed data into the FOA-GRNN model for simulation,and compare it with other commonly used neural network methods.The results show that the newly established model is compared with the models established by other commonly used 4 neural network methods.The predictive ability of FOA-GRNN is the best,and the predictive ability of the model built using traditional methods is slightly inferior,which proves that FOA-GRNN There is greater potential in the direction of rolling force prediction.