Research On Multi-modal Process Monitoring Method For Continuous Annealing Furnace

With the development of industrial technology,the requirement of the product is ever-growing,the scale and complexity of modern industrial systems are also on the rise.Annealing furnace equipment in the hot-rolled process occupies a very important position,it's operating conditions can have a direct impact on product quality and energy consumption.Because of its own complex characteristics,it is impossible to monitor the running state of the annealing furnace directly using simple monitoring methods.Therefore,the study of annealing furnace multi-modal operation monitoring method becomes very important.Multivariate statistical methods are widely used in process monitoring,It mainly realizes the on-line monitoring of running process by analyzing and interpreting the data set of process variables.In this paper,after reading a large amount of literature,the following research is mainly carried out in the following aspects:(1)Research on process monitoring method in single mode.After theoretical analysis and simulation research,It is found that KECA has some advantages over KPCA in principal component metadata extraction.Aiming at the problem of industrial process data accompanied by noise,with KECA as the main algorithm,wavelet transform is introduced to process the data to form an effective monitoring method of wt-keca single modal process.(2)To analyze the multi-modal characteristics of the continuous annealing process,a multi-modal method is proposed.Mainly includes two aspects:one is the division of steady state and steady state,the other is the division of transition state between steady states.The PSO-FCM method is used to data clustering,and the transient information is divided according to the variable rate of change.(3)In order to improve the monitoring method for different modalities,different fault types have better monitoring capabilities,based on bayesian inference,the wt-keca method is integrated to solve the problem of kernel function parameters.Furthermore,the multi-modal process monitoring simulation study of annealing furnace is carried out in the background of continuous annealing furnace.Based on the simulation results of the annealing furnace,the KPCA method is basically effective for process monitoring,but its dimensionality principle leads to the ambiguity of the contribution rate of variable principal components.With the same monitoring effect,KECA method becomes clear on the analysis of variable contribution rate.The method of WT-KECA integrated by bayes achieves the ideal effect on the monitoring of continuous annealing.