Theoretical Study On Thickness Quality Defect Tracing Of Hot Rolled Aluminum Alloy Strip

With the development of traditional industry towards the direction of lightweight and economy,the demand of aluminum alloy strip is increasing,and the quality requirements of aluminum alloy strip are becoming more and more strict.As an important process of strip forming,hot rolling has an important influence on the quality of aluminum alloy.The longitudinal thickness is one of the important indexes to evaluate the quality of aluminum alloy sheet and strip.For aluminum alloy sheet and strip used under severe working conditions,the requirement of longitudinal thickness is more strict.However,it is still a difficult problem to find out the technological factors leading to the quality defects of strip thickness when the longitudinal thickness is out of tolerance.Therefore,the main breakthrough direction of intelligent and high-precision production of aluminum alloy sheet and strip is to realize the pre judgment of the thickness quality of aluminum alloy sheet and strip before production,the monitoring during production and the traceability after production.Based on the research difficulties of aluminum alloy thickness quality traceability,this paper establishes the quality monitoring and quality traceability diagnosis model for the whole process of aluminum alloy plate and strip hot rolling.In this paper,the status quo of quality traceability diagnosis of plate and strip thickness and the research status of some intelligent algorithms in hot strip quality diagnosis are discussed,especially the research direction based on data-driven.In view of the blind spot of current research,a research scheme of adopting data-driven intelligent algorithm based on the formation mechanism of plate thickness is proposed.Combined with the actual production process of aluminum alloy plate and strip in a production line,the parameters related to thickness and quality are sorted out as the data source of the follow-up data model.The parameter monitoring of single stand is far from meeting the quality monitoring requirements of the whole process.Therefore,based on the process model of single stand,a monitoring method is proposed to establish the whole process monitoring model by using the "hidden variables" of single stand,so as to realize the layered and block monitoring of the whole hot rolling process.The whole process monitoring model can not only integrate the sub process monitoring model,but also reflect the abnormal matching between sub process models.The sub process and the whole process model can accurately and timely monitor the abnormal parameters.Based on the thickness quality monitoring model of aluminum alloy hot rolling,a Bayesian(IBN)traceability model incorporating importance factor is established,which solves the problem that it is difficult to quantify the conditional probability between parameter nodes in the hot rolling process,and improves the prior node probability shift caused by inaccurate learning of traditional Bayesian parameters.The results show that the traceability accuracy of IBN is 80%,which has a certain guiding significance for the quality traceability of actual production.