论文部分内容阅读
针对基于支持向量机(Support Vector Machine,SVM)的间歇过程故障诊断准确率低的问题,结合间歇过程的时段特性,提出了一种基于子时段MPCA-SVM的间歇过程在线故障诊断方法。首先,利用多向主成分分析(Multi-way principal component analysis,MPCA)提取出间歇过程正常运行状态下的每个采样点的主成分,将相邻的且具有相同主成分个数的采样点归到同一粗划分时段内,再在每一个粗时段内利用相邻采样点的负载矩阵的角度信息作为相似性判据来细化分时段;其次,对每个时段建立MPCA在线过程监测模型,同时,利用MPCA提取每个时段内各个类型故障的特征,并用特征数据建立SVM故障诊断模型;最后,MPCA监测模型实施监测功能,当检测到故障时,相应时段的SVM故障诊断模型进行诊断。将该方法应用于青霉素发酵过程仿真平台进行验证,该方法相比于不分时段的SVM的故障诊断方法,平均可提高故障诊断准确率11%,实验结果表明了该方法的有效性和可行性。
In order to solve the problem of intermittent process fault diagnosis based on Support Vector Machine (SVM) with low accuracy, combined with the time characteristics of batch process, an online fault diagnosis method based on sub-period MPCA-SVM is proposed. First, the principal component of each sampling point in the normal operation of the batch process is extracted by using Multi-way principal component analysis (MPCA), and the adjacent sampling points with the same main component number In the same rough dividing period, the angle information of the adjacent sampling points is used as the similarity criterion to refine the sub-periods in each rough time. Secondly, the MPCA on-line process monitoring model is established for each period, meanwhile, , The MPCA is used to extract the features of each type of fault in each time period and the SVM fault diagnosis model is established by using the feature data. Finally, the MPCA monitoring model implements the monitoring function, and when the fault is detected, the SVM fault diagnosis model of the corresponding period is diagnosed. The method is applied to the simulation platform of penicillin fermentation process for verification. Compared with the SVM fault diagnosis method, the method can improve the accuracy of fault diagnosis by 11% on average, and the experimental results show the effectiveness and feasibility of the method .