Research On Methods And Application Of Fault Diagnosis Based On The Kernel Method

Along with the application of automation technology in steel and iron and non ferrous metal industry in our country, the request of reliability and security of rolling control process is getting higher and product quality is becoming stricter. The fault diagnosis of strip mill has become one of the most important research directions in the metallurgical automation field. The statistical method based on the data is more practical, for this method doesn’t depend on precise mathematical model and the on-line process data of industrial production are easy to obtain. This method need to collect the data both under the normal condition and fault condition. The diagnosis process need to include fault detection and fault identify.The thesis draws kernel method into traditional Principal Component Analysis (PCA) and Kernel Fisher discriminant analysis (KFDA), and then researches the methods and application of fault diagnosis based on the KPCA and KFDA. The fault diagnosis need to include fault detection and fault identify. The first step is fault detection. The thesis adopts the methods based on the PCA and KPCA through their statistics of Tz and SPE. The second step is fault identify. The thesis adopts the methods based on the FDA and KFDA through the design of multiple-valued classifier. To reduce run-time classifier complexity and improve the classification accuracy, it’s requred to remove the correlation between raw data and to reduce the dimension of the original data. The thesis adopts the KPCA method to preprocess the original data. In view of the kernel function parameters selection influences the performance of multiple-valued classifier based on the KFDA method greatly, the thesis utilizes the PSO algorithm to optimize parameters of kernel function of the multiple-valued classifier. The thesis gives out the optimization KFDA algorithm based on the PSO algorithm that utilizes the class separability as the evaluation index. Combining the optimization KFDA algorithm with feature extraction based on the KPCA method, the thesis obtains that the accuracy of fault diagnosis can reach more than90%through the simulation results. The results demonstrate the effectiveness of the method.