中文摘要：

针对传统多变量统计过程控制（MSPC）方法在故障检测、故障原因分析和故障识别中的难点，提出了多元特征提取方法与基于支持向量机（SVM）的一类分类器设计、特征选择以及多类分类器设计方法相结合的一种完整的MSPC新方法．该方法在故障检测中可去除特征满足特定分布的假设前提。并可确定多个统计量的控制限；在故障原因分析中综合考虑故障对于各个变量大小的影响以及变量变化对于故障分类的重要性，提高了关键变量选择的准确性；并且故障识别是基于SVM对故障特征分类的优良特性，避免了传统判别法中经验准则的引入．上述方法在标准仿真平台Tennessee Eastman过程上结合主元分析（PCA）进行了应用，结果显示了其优越性．

英文摘要：

In multivariate statistical process control（MSPC）, fault detection, fault identification and fault isolation are challenging problems. An integrated novel MSPC method was proposed by combining multivariate feature extraction with three SVM-based methods commonly used in one-class classifier design, key feature selection and multi-class classifier design, respectively. In this method it was able to calculate control limits of multiple statistics for fault detection simultaneously without conventional theoretical distribution assumptions, and to determine the key variables for fault identification based on both their magnitude changes and their contributions to fault classification in residual space which improves the identification accuracy, and fault isolation was implemented by taking advantage of the well-known properties of SVM-based multi-class classifier which avoids introducing specific discriminant criteria. Using principal component analysis （PCA） as feature extraction method, the SVM-based MSPC method was demonstrated by application to a benchmark simulator Tennessee Eastman process. The results showed the advantages.