Analysis And Optimization Of Hot Strip Shape Based On Data Driven

The iron and steel industry is the pillar industry of the Inner Mongolia Autonomous Region.The steel rolling process is an important link in the production of steel plate and strip steel.The continuous and efficient operation of the production process is an effective way to enhance the competitiveness of enterprises.The quality of the plate shape in the production process of plate and strip steel directly affects the economic value and application value of the plate.Because the hot tandem rolling production process is a complex production process with multiple parameters,strong coupling,multiple interference,nonlinearity,and rapidity,the plate and plate Shape control becomes very difficult.This paper takes the key processes of a 2250 mm hot strip production line in a plate factory as the research object.In order to further improve the plate shape quality,based on the big data information of the production process,the research on data intelligent analysis and key process parameter optimization decision-making is carried out,including the following Several aspects of research:(1)A detailed mechanism analysis was carried out on the main setting parameters(rolling force,plate width)that affect the plate shape quality in the hot strip production.At the same time,the shape control system was analyzed,and the data-driven rolling was determined.Optimization method of braking force and board width setting value.(2)Analyze the data characteristics of the existing production process,preprocess the data,use the isolated forest algorithm to process the outliers of the production data,improve the modeling accuracy of the optimization model;use the maximum mutual information coefficient method to determine the plate width and rolling The braking force setting value optimizes the input parameters of the model,and improves the training and prediction speed of the model while ensuring the accuracy of the model.(3)In the plate shape setting,in view of the deviation of the plate width setting value,the Light GBM algorithm is used to optimize the plate width setting value,and the random search algorithm is used to optimize the hyperparameters of the model.Experimental results show that the prediction error of Light GBM board width prediction algorithm has a hit rate of 92.22%within ±2mm,which is smaller than the prediction error of support vector regression,multilayer perceptron and XGBoost board width prediction algorithm.The prediction result of Light GBM plate width prediction algorithm is closer to the actual strip width,and the optimized plate width setting value is used to replace the calculated value of the mechanism model to improve the rolling force setting accuracy and the bending roll calculated by the plate shape setting model.The setting accuracy of force and roll shifting,L1 level using the optimized plate width setting value can improve the plate shape quality of the output plate and strip steel products.(4)In the flatness setting,in view of the deviation problem of the setting value of the rolling force,the random forest regression algorithm is used to optimize the setting value of the rolling force.The particle swarm algorithm is used to optimize the hyperparameters in the random forest regression algorithm to ensure that the random forest regression algorithm will not reduce the model accuracy due to non-optimal parameters.The experimental results show that the rolling force prediction algorithm of random forest regression has smaller prediction errors and shorter prediction time than support vector regression,multilayer perceptron and radial basis neural network rolling force prediction algorithms.Compared with the calculated result of the mechanism model,the predicted result has a smaller deviation from the actual measured rolling force.The optimized rolling force setting value is used to replace the calculated value of the mechanism model to increase the roll bending force and roll shift calculated by the flatness setting model.L1 level uses the optimized setting value to improve the shape quality of the output plate and strip steel products.