CPCA 监视与知识诊断相结合的智能诊断

将多元统计分析技术与基于知识的故障诊断方法相结合, 提出了一种过程智能监视与故障诊断方法. 利用多元统计一致主元分析 (CPCA) 方法对过程异常工况进行监视, 以找到故障发生区域及发生时段; 针对故障发生区域, 将 CPCA 方法得到的整体及局部的统计指标的定量信息、统计指标超限与否的定性信息以及得分及贡献率等综合信息提供给故障区域的基于知识的专家系统. 从连退过程的应用可看出, 该方法可检测出单变量监视难以检测的故障, 并得到可靠的故障诊断结论.

將多元統計分析技術與基于知識的故障診斷方法相結合, 提出了一種過程智能監視與故障診斷方法. 利用多元統計一致主元分析 (CPCA)方法對過程異常工況進行監視, 以找到故障發生區域及發生時段; 針對故障發生區域, 將 CPCA 方法得到的整體及局部的統計指標的定量信息、統計指標超限與否的定性信息以及得分及貢獻率等綜合信息提供給故障區域的基于知識的專家系統. 從連退過程的應用可看出, 該方法可檢測出單變量監視難以檢測的故障, 并得到可靠的故障診斷結論.

Combining the techniques of multivariate statistical process monitoring and knowledge based fault diagnosis, an intelligent monitoring and fault diagnosis method for industrial process is proposed in this paper. The multivariate statistical consensus principal component analysis (CPCA) method is adopted to monitor the process, and to determine where and when the fault occurs. Moreover, the whole and regional statistical indices quantitative information as well as qualitative information, and the scores and contribution plot information are obtained by the CPCA method, and provided to the regional knowledge-based expert-system. Thus, from the application of this method to the continuous annealing process, the proper monitoring results which could not be achieved by the univariate monitoring and the reliable diagnosis conclusions can be achieved.