基于PCA-SVM的储运过程故障诊断方法  

Fault Diagnosis Method during Storage and Transportation Process Based on Principal Component Analysis and Support Vector Machine

作　　者：房汉鸣 税爱社[1] 卿宇搏 宗福兴[3] Fang Han-ming;Shui Ai-she;Qing Yu-bo;Zong Fu-xing(Dept. of Logistics Information & Logistics Engineering;Dept. of Management Science & Engineering,LEU,Chongqing 401311,China;Unit 76174,Shaoguan Guangdong 512000,China)

机构地区：[1]后勤工程学院后勤信息与军事物流工程系 [2]76174部队 [3]后勤工程学院管理科学与工程系

出　　处：《后勤工程学院学报》2016年第4期86-91,96,共7页Journal of Logistical Engineering University

摘　　要：针对储运过程工艺复杂、监控变量多且故障样本数据相对有限的问题,在介绍主元分析原理和支持向量机方法的基础上,提出了主元分析与支持向量机相结合的储运过程故障诊断方法,建立了提高故障诊断速度和诊断性能的故障诊断模型。首先采用主元分析法进行特征提取实现降维,其次构造新的训练和测试样本集,最后训练支持向量机分类器的故障诊断流程。以数字化油库为仿真实验对象,进行了油料收发过程中的故障诊断实验,验证了所提方法的有效性。Aimed at the complexity of storage and transportation process work flow, the multidimensionality of monitoring andcontrol parameters and the small property of fault mode, the combined fault diagnosis method that combined principal componentanalysis with support vector machine was put forward on the basis of introducing the principle of principal component analysis andsupport vector machine. The fault diagnosis model which could improve the speed and performance of fault diagnosis. The fault diag-nosis flow was given, in which PCA was first used for feature extraction and realization of dimensionality reduction, new training andtest sample sets were built, based on the calculation results secondly, and the support vector machine classifier was trained last. Thepaper took the digital terminal as the simulation object and conducted the fault diagnosis experiment during the process of oil tran-sit. The effectiveness of the proposed method was verified.

关 键 词：主元分析 支持向量机 储运过程 故障诊断 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置]