Research On Optimization Of Strip Cooling Process Control Based On Data Driven

Currently,the demand for steel products in various industries is also increasing,an important issue in the steel industry is how to further improve the quality of products.The laminar cooling of the strip after hot rolling is the last key link of the hot rolling production line.Its main role is to cool the strip after hot rolling,so that the temperature of the strip after cooling reaches the tolerance range allowed by the target coiling temperature.The control effect of the strip temperature after cooling seriously affects the tissue properties of hot rolled strip.In order to ensure the high quality and quantity of the finished strip,it is imperative to study the optimization of the operation of the strip cooling process after hot rolling.In actual production,due to errors in strip race division,changes in cooling environment or water volume,resulting in a weak head setting capability for the first strip after in switching steel races or a long rolling stop,which seriously restricts the high precision control of the strip full-length temperature.In this thesis,the laminar cooling part of the 2250 mm hot rolling production line in B steel plant is taken as the research object.In view of the problem of weak head setting ability of the first strip rolling after switching steel races or long time stopping rolling in this line,the combination of industrial big data,expert knowledge and intelligent models optimizes the strip cooling process after hot rolling.The main research work and achievements are as follows:(1)Problem description and solution study of strip cooling process after hot rolling.Familiar with the development of steel hot rolling cooling technology and control technology difficulties,detailed description of the laminar flow cooling process process and coiling temperature control strategy.In view of the existing problems of this production line,an intelligent control optimization scheme combining industrial big data,expert knowledge and intelligent model is proposed to improve the control accuracy of the full length temperature by modifying the set opening number of valves.(2)Laminar flow cooling process data processing.The input variables affecting the coiling temperature prediction model and the header valve opening number setting model were determined by combining the actual production process and expert experience.Aiming at the problem that the laminar cooling process data set collected from the field contains a lot of noise,data processing is performed on the data set,including outlier detection,missing value repair,standardization.After the above data processing operations,high-quality modeling data sets are obtained,including 11 000 sets of data sets related to the coiling temperature model and 21 400 sets of data sets related to the valve opening number setting model.(3)Coiling temperature prediction modeling.Establishing coiling temperature prediction models based on gbdt,svm,wnn and gbdt,svm,wnn optimized by differential evolutionary algorithm using relevant high-quality modeling data sets,and judging the performance of the above six prediction models through several model evaluation indexes.The experimental results show that the prediction model based on the differential evolution algorithm to optimize the gradient boosting decision tree has better performance in the prediction of coiling temperature,and its various error indicators are the smallest,and the prediction hit rate is the highest,reaching98.5%.(4)Optimization study of valve opening number setting based on Bayesian optimization.The coiling temperature of the strip steel head is estimated by the coiling temperature prediction model,and the status of the strip cooling production process is judged according to the expert experience.If the deviation between the model prediction and the target is too large,the valve opening number setting model based on the Bayesian optimization gradient boosting decision tree is used to correct the set opening number of the cooling zone valves.The number of valve openings in the cooling zone can be reduced for the overcooled strip,and the number of valve openings can be increased for the undercooled strip,which can improve the control hit rate of the strip coiling temperature.