The Identification And Intelligent Forecast Of Hot Rolled Thickness

For plate steel, the overall dimensions including width, thickness, strip shape and platecrown, plane shape and so on. In recent years, with the development of manufacturing industry,metallurgical industries pay more attention to the quality of strip, especially to thethickness precision index. Therefore, the thickness control precision problem has a growinglimitation in the improvement of the product quality. Improving the predictive accuracy of platestrip thickness plays a crucial role in plate strip production. There are two aspects in thetechnology of solving the problem of thickness prediction. One is the dealing with the analysisof historical data, digging out the useful information. The second one is the establishment of theprediction model. Emphatically discuss the two aspects in research and present situation ofapplication both at home and abroad and the main problems existing, and then points out theresearch status and development trend in this field. The main research work is as follows:This paper expounds the development status of strip steel thickness prediction for hotcontinuous rolling and carried on a detailed analysis and research on the mechanism offinishing mill group of thickness distribution model, the cause of plate strip thicknessfluctuation, thickness variation law and made a concrete analysis about the present situation ofmechanism of the model prediction. The paper summarizes the main factors influencing thethickness of strip steel exports for hot continuous rolling on the basis of the above analysis:finishing the entry thickness, roughing exit the measured temperature, roll gap, rolling force,the seventh frame actual entry thickness, strip temperature, rolling speed, etc.In view of the large number of detailed data in the actual rolling production, principalcomponent analysis and rough set are used to analyze and process it. It is proved to be verifiedof the two methods in dealing with the continuous strip data from the point of processing result.Analyze the main factors influencing the finishing exit thickness including: the seventh frameentry thickness, roll gap, entry thickness and rolling force.On the basis of the data processing result, RBF neural network prediction model isestablished, the neural network toolbox in MATLAB is used to simulate and forecast. Thesimulation results show that the prediction effect of the model of RBF neural network issuperior to that of the mechanism model. In consideration of the good prediction effect and wide application of the combinationforecasting model, establish a combination forecast model of the strip thickness atthe exit of stand in the hot rolling based on the GRNN. In the practice, single forecasting modelis selected including RBF neural network prediction model, BP neural network predictionmodel and the SVM prediction model. The experimental results show that the non-linearcombination of the forecasting method based on GRNN can improve the prediction accuracyeffectively.Establish a prediction model close to the actual production of the strip steel though theresearch. The paper has a certain value in improving the control accuracy of plate thickness ofstrip steel.