Heating Process Monitoring Of Reheating Furnace Based On Improved Multi-Way Kernel Entropy Component Analysis Method

With the development of industrial technology,the requirement of the product is ever-growing.The reheating furnace is a very important equipment in the hot rolling process.Product quality and energy consumption can be affected by the operating state of the reheating furnace.However,the simple monitoring method can not be used to monitor the running state of the reheating furnace directly,due to the complex characteristics of reheating furnace.Therefore,the research on the monitoring method of the reheating furnace is very important.Process monitoring method based on multivariate statistical is a very important branch in the research of process monitoring and fault detection field.On-line monitoring and identification of the abnormal situation can be carried out through the analysis and interpretation of the data sets of the process variables.In this paper,the running state of reheating furnace is monitored by multivariate statistical method.Details are as follows:(1)Due to the change of rolling rhythm,the time in furnace of all batches billet can not be completely same,which will make the data length of different batches unequal,so the sampled data can not form a balanced 3D data matrix.To solve this problem,the DTW method is used to process the data before building the monitor model.This method can be optimized by using weighted coefficient and window when it is used for processing data,which can also solve the problem of data ill structured when using DTW.(2)In order to solve the nonlinear problem in the reheating furnace,the multiway kernel principal component analysis(MKPCA)and the multiway kernel entropy component analysis(MKECA)are used to monitor and analyze the state of the reheating furnace.(3)Moreover,the industrial process is dynamic,the variables are usually auto-correlated,canonical variate analysis(CVA)and multiway kernel entropy component analysis(MKECA)are combined to solve the problem of auto-correlation.The monitoring simulation of reheating furnace shows that MPCA method can monitor some cases,but the false alarm rate and the detection rate are both not ideal.MKPCA and MKECA hava a better performace than MPCA on fault detection rate,however they do not have a good performance on false alarm rate.While MKECA combined with CVA can have a great advantage both on false alarm rate and the detection rate.The effectiveness of the improved MKEC A is demonstrated by simulation results...