Research And Implementation Of Strip Thickness Prediction System Based On Improved HGWO-SVR Model

With the development of industry and the progress of science and technology in China,more and more traditional industries are combined with modern science and technology to develop refined and automatic management.At the same time,the products that industry provides for people,both in terms of performance or function,have been improved.Therefore,higher quality requirements are put forward for the most widely used steel in industrial products.In the process of strip rolling,the control of strip thickness accuracy is the key to improve its quality.However,due to the complex process,there are many factors that affect the thickness of strip,and different factors have different influence on the thickness of strip,and the influence of each factor is nonlinear.Therefore,if want to export high-quality strip,the intelligent control of thickness accuracy is an urgent problem to be solved.The traditional control method can not meet the accuracy demand of strip thickness.With the continuous development of artificial intelligence in various industries,intelligent control of rolling has become an important direction of artificial intelligence research.Base on machine learning theory,this paper designs an improved gray wolf algorithm to optimize the support vector regression(HGWO-SVR)model for strip thickness prediction system.Due to the influence of penalty factor P and kernel function parameter ? when the traditional SVR model is used for regression prediction,the prediction result is not ideal.In this paper,gray wolf algorithm is used to optimize the parameters P and ? in SVR model,but because of the problem that the initial population is easy to fall into local optimum and the global search ability is weak in the later stage,differential evolution algorithm is used to keep the diversity of gray wolf population,and an improved differential gray wolf algorithm is proposed to improve the prediction accuracy of the original gray wolf algorithm.Finally,the improved HGWO-SVR prediction model is applied in the industrial strip rolling to predict the strip thickness.The system uses the actual data generated in the rolling process of a domesticsteel group factory as the sample set of the prediction model.The system is divided into the following five modules: information input module,system management module,data management module,prediction model building module,strip thickness prediction module.In the data management module,the mutual information calculation method is used to select the features that have a great impact on the thickness of the plate and strip,and the selected parameter data is standardized as the input of the model.The prediction model building module uses the processed training set data,and uses the improved HGWO algorithm to optimize the parameters in SVR model,and constructs the prediction model.Using the improved HGWO-SVR model to predict the strip thickness can effectively improve the prediction accuracy.Through repeated tests,the prediction results of the prediction system designed in this paper meet the quality requirements of strip rolling in terms of thickness,and the system meets the expected standards in function and performance.