Quality Monitoring Of Strip Thickness And Shape And Identification Of Abnormal Information Transmission Path In Hot Strip Rolling

The hot rolling production process is continuously produced by multi-stand coordina-tion.The nonlinear coupling of parameters in the rolling unit and the cumulative inheritance of quality between units cause problems such as difficult positioning of quality-related faults in the process industry,difficult identification of abnormal information transmission paths,and easy evolution of faults.To effectively ensure product quality and improve the ability of quality-related fault monitoring and diagnosis,this paper focuses on the quality control requirements of steel manufacturing process.Taking the strip quality of 1580 seven-stand hot continuous rolling process in a factory as the research object,based on industrial data,the accurate diagnosis of quality-related faults in rolling process and the identification of abnormal information transmission path are studied.Through the theory and method of roll-ing mechanism model and multivariate statistical algorithm,the research on the influence characteristics of multi process unit variables on quality,the quality related fault monitoring theory of hot rolled strip and the identification method of quality abnormal information transmission path are mainly carried out.The research results have important theoretical significance and engineering value for revealing the essence of product quality correlation of hot rolled strip,improving the level of product quality related fault diagnosis of multi-process units of hot rolled production and the intelligent level of hot rolled production.In the multi-stand rolling process,the contribution of each stand variable to the quality is different.However,the diagnosis process based on the traditional data model ignores the difference of the influence of process variables on the quality,and the lack of explanation of the variable mechanism causes the misjudgment of the cause of the quality anomaly.To ensure the accuracy of the diagnosis of the causes of quality anomalies,this paper optimizes the data model by weighting the mechanism weight,improves the mechanism interpretation of the variables in the diagnosis model,reveals the contribution of each variable in the di-agnosis process to the quality,and provides a precise model basis for revealing the infor-mation transmission relationship of quality anomalies.In this paper,a KPLS quality moni-toring and diagnosis method based on distributed strategy is proposed.By integrating the distributed idea and according to the characteristics of multi-stand production,it is divided into sub-module systems with single rack as the smallest unit to carry out quality-related fault monitoring and diagnosis.The fluctuation of process variables of each subsystem is monitored by T~2and SPE statistics,and the contribution of each subsystem variable is con-structed by nonlinear contribution diagram to realize the identification of quality-related abnormal variables.There is information exchange and quality information transmission between the hot rolling process submodule system and its adjacent modules.In order to reveal the quality information transmission path and identify the causes of abnormal strip steel quality,an asymmetric transfer entropy matrix is introduced to construct a directed transfer topology structure with variable causality,a WKPLS-MIC-TE method for diagnosing thickness anomalies and identifying information transmission paths of hot rolled strip is proposed based on distributed monitoring and diagnosis strategies.The transmission path identifica-tion of quality related information in multi-stand hot rolling stage is realized.In the face of shape quality,the data is a kind of surface data characteristics.Aiming at the multi-dimensional characteristics of the shape quality data,the shape characteristics are parameterized and the quality evaluation of the shape is realized by accurately predicting the shape characteristics.At the same time,on the basis of accurate prediction of flatness quality characteristics,a hierarchical path identification method for flatness quality infor-mation transmission based on BN-e IWOA is constructed.The Bayesian network is opti-mized by the improved whale optimization algorithm,and the optimal information trans-mission network topology of each stand module variable in the rolling process is constructed.It provides a new and effective method for the diagnosis of strip quality and the traceability of abnormal causes with multi-dimensional characteristics,to carry out information feed-back in advance when the shape has a large deviation,and improve the control ability of the hot rolling process to the shape.