关键词: 动态过程, 质量异常模式, 多主元特征, 支持向量机

Abstract: To improve the recognition accuracy and reduce the computation complexity in dynamic process,a novel quality abnormal monitoring model based on Multiple Principal Component analysis integrated Support Vector Machine (MPC-SVM) was proposed.The different PC vectors were extracted from original data matrix by PCA technique,and the extracted principal components were taken as the input of SVM classifiers to train the model.After integrating transform matrix of PCA with MSVM,MPC-SVM recognition model for quality abnormal patterns in dynamic process was established.Compared with recognition model based on other traditional methods,the simulations illustrated that the proposed MPCA-SVM model had a good performance than other models in the fields of train time and recognition accuracy.

Key words: dynamic process, quality abnormal monitoring, multiple principal component, support vector machine