Research And Design Of Strip Width Prediction System In Hot Continuous Rolling Mill

Achieving high-precision strip production in hot-rolled strip rolling production lines has always been one of the "Made in China 2025" industrialization strategies.Most of the width control process of the hot rolling production line is in the rough rolling area.The control system is affected by on-site factors such as roll wear and equipment vibration,resulting in the reduction of the rolling width accuracy of the billet in the entire rough rolling line.Improving the prediction model of strip width in roughing mill is an effective solution to solve the prediction accuracy of strip width in the hot rolling mill.However,the current fixed widening model relies on manual experience,and there are many unreasonable factors.To solve the problem of low prediction accuracy of the strip width in the hot-rolled strip production line,this paper analyzes the strip width control principle of the roughing mill.Combined with a large amount of historical operation data,a hybrid modeling method combining mechanism and data is proposed.A strip rolling production line of a hot-rolling plant is taken as the research object to establish a hot-rolled strip width prediction model.The specific research contents of this paper are as follows.(1)In this paper,an in-depth analysis of the strip width control mechanism during the operation of the hot rolling mill is carried out.At the same time,the setting model of vertical roll pressure,vertical roll reduction setting model,target width setting model of the rough rolling mill are analyzed.Given the influence of many environmental factors in the industrial site,the factors affecting the width accuracy of strip steel products are objectively analyzed.Aiming at the defects of the current strip width model,to improve the width accuracy of strip products,an improved idea based on the target width setting model is proposed.(2)Aiming at the problem of large fluctuation and abnormal value of width data collected by hot rolling field sensors,the upper and lower dispersion coefficients combined with wavelet analysis are used to preprocess the data to eliminate abnormal values.For the problem of different dimensions of feature parameters,the Z-Score standardization method is used to standardize the data to eliminate the influence of different dimensions.The preprocessed data is stored in the rough rolling width pre-calculation database to provide a reliable sample data set for training the width prediction model.(3)Based on the pre-calculation database of strip width in rough rolling mills,the research on the width prediction model based on the data-driven modeling method is carried out.After the mechanism analysis,the characteristic variables that have a great influence on the strip width at the rough rolling exit are initially obtained,and the variables with high Pearson correlation coefficient are selected as the input of the width prediction model,and the strip width prediction model based on particle swarm optimization least squares support vector machine is established.After simulation verification,the proposed model meets the prediction accuracy requirements.(4)To improve the prediction accuracy of the strip width calculation model of the hot rolling roughing mill,this paper uses the differential evolution algorithm with improved mutation strategy to conduct parameter search;and establishes a width prediction model based on the optimized width coefficient of Takashi Shibahara.On this basis,to make the width model more accurate,a hybrid prediction system for strip width is jointly established using parallel data-driven models;and designed a visual interface.After simulation verification,the proposed prediction system meets the prediction accuracy requirements and the industrial field fitting positive tolerance requirements.