Research On Deep Optimization Of Process Parameters Of Hot Strip Mill Based On Data Driven

With the development of high-tech and steel industry,the application of plate and strip is becoming more and more extensive,and at the same time,higher requirements are put forward for the production of plate and strip.However,the vibration of the rolling mill will lead to poor stability of the rolling mill,and it will also cause light and dark vibration lines on the surface of the strip,wear and tear of the roll and even fall off,resulting in damage to the rolling mill equipment,which greatly affects the production quality of the strip and the personal safety of the on-site staff.It has affected the production efficiency and caused immeasurable harm to the enterprise.Therefore,it is of great research significance to propose effective suppression measures for rolling mill vibration.This paper takes F2 frame and F3 frame of a hot continuous rolling mill as the research object.Aiming at the severe vibration problem of F2 frame and F3 frame,the vibration data and process parameter data of the on-site rolling mill are tracked and collected.The vibration and process parameters of F2 and F3 stands are analyzed,and it is concluded that F2 and F3 stands have the same rolling mill vibration form,but the vibration intensity of F2 stand is stronger than that of F3 stand.The reason is that the vibration is transmitted from the F2 mill to the F3 mill through the strip(thickness fluctuation,bright and dark vibration lines).Therefore,the F2 frame and the F3 frame are considered jointly,and the optimization of process parameters and the suppression of vibration are considered together.The pass value of the rolling mill is used as the input and output of the data model.Therefore,through the time-frequency analysis of the vibration signal,the effective value of the vibration is selected to represent the pass value of the vibration,and the average value of the process parameters is selected to represent the pass value of the process parameter.Then the abnormal value of the collected data is processed,and the data is normalized.Secondly,with rolling process parameters as input and vibration as output,a random forest vibration prediction model and ELM-Adaboost vibration prediction model were established.The comparison found that the prediction accuracy and result stability of the ELM-Adaboost vibration prediction model were significantly better than Random forest predictive vibration model.Therefore,the ELMAdaboost prediction vibration model is used as the fitness function of the SPEA2 optimization algorithm,the optimization model is established with the vibration value of the rolling mill as the optimization target,and the optimized characteristic variables are selected as the rolling process parameters for iterative optimization.Finally,the ELM-Adaboost vibration prediction model is used to quantitatively analyze the relationship between process parameters and rolling mill vibration.Predict by modeling the F3 stand and F3 stand data,and then use the SPEA2 algorithm to optimize the process parameters.The results show that the optimized rolling mill rolling force,front tension,and rolling speed are significantly reduced,and the back tension is increased..The optimization results are consistent with the quantitative analysis results,which verifies the feasibility of process parameter optimization for the suppression of rolling mill vibration,effectively reduces the vibration of F2 stand and F3 stand,and improves the operation stability of the rolling mill.