【摘要】 为克服传统过程监控方法需假设过程特征信号服从多元正态分布的缺陷,本文提出了一种将独立成分分析(ICA)与支持向量机结合的故障诊断方法。通过建立独立成分模型确定相应的统计量界限,筛选出需进一步检测的故障数据,再由支持向量机进行故障识别。将该方法用于化工聚合反应的过程监控与故障诊断中,仿真结果表明,这种混合故障诊断方法通过适当地调节统计量控制界限,不仅能够正确识别故障,而且能够纠正由误检数据引起的误报,提高故障诊断的准确率。更多还原

【Abstract】 In order to overcome the shortcoming of the conventional process monitoring method’s assumption that the extracted features must be subject to multivariate normal distribution,a novel method of fault diagnosis combining with Independent Component Analysis (ICA)and Support Vector Machines(SVM)is presented.The fault data detected is determined by the bound of corresponsive statistic in the Independent Component Model firstly,then the failure category is identified by Support Vector Machines(SVM).This hybrid method is applied to a system of process monitor and fault analysis for a chemical polymerization reaction process.The simulation result shows that the hybrid method of ICA and SVM not only can make accurate fault recognition,but also rectify the false alarms proceeded from the mistaken data by adjusting the control bound of process statistics.Therefore,this hybrid method can increase the accuracy of fault diagnosis.更多还原