基于支持向量机的造纸废水处理过程故障诊断

Fault Diagnosis of Papermaking Wastewater Treatment Processes Based on Support Vector Machine

中文关键词:  故障检测  故障诊断  主成分分析  马氏距离判别分析  支持向量机

Key Words:fault detection  fault diagnosis  principal component analysis  Mahalanobis distance discriminant analysis  support vector machine

中文摘要:

      故障检测和故障诊断是工业过程监控的主要内容。针对造纸废水处理过程的多变量、非线性、大时变等特点，本课题首先采用主成分分析（PCA）对故障进行检测，然后分别采用马氏距离判别分析和支持向量机(SVM)对偏移、漂移和精度下降3种故障类型进行故障诊断。计算结果表明，基于主成分分析的故障检测率达97.50%；基于支持向量机故障诊断方法的故障分离能力为90.00%，而基于马氏距离判别分析方法的故障分离能力为73.75%。相比基于马氏距离判别分析的故障诊断方法，基于支持向量机的故障诊断方法更适合于非线性时变的造纸废水处理过程。

Abstract:

      Fault detection and fault diagnosis are main topics in the industrial process monitoring field. Concerning the complicated character-ristics of papermaking wastewater treatment processes (WWTP), principal component analysis (PCA) was used for the fault detection, firsthy. Then Mahalanobis distance discriminant analysis and support vector machine (SVM) were used for the fault diagnosis of three constructed types of sensor faults, respectively. The results showed that the fault detection rate using PCA was 97.50%. The separation rates using SVM and Mahalanobis distance discriminant analysis were 88.75% and 76.25%, respectively. Compared with the fault diagnosis method based on Mahalanobis distance discriminant analysis, the fault diagnosis method based on SVM was more suitable for the papermaking WWTP.