基于支持向量机与知识的汽轮发电机组智能故障诊断研究  

Study of Intelligent Fault Diagnosis Method for Turbo-generator Unit Based on Support Vector Machine and Knowledge

作　　者：黄乃成[1] 顾煜炯[1] 谢骐宇[1] 张国坤[1] 

机构地区：[1]华北电力大学电站设备状态监测与控制教育部重点实验室,北京102206

出　　处：《现代电力》2012年第6期68-73,共6页Modern Electric Power

基　　金：中央高校基本科研业务费专项资金项目(12QX06);华能集团科技创新项目(HNKJ11-H27)

摘　　要：针对汽轮机组故障诊断效率与准确率不高的问题,把常见的18种振动故障划分为4类,利用支持向量机对振动信号的频谱进行故障类识别,实现故障的初诊断。对不同类故障建立不同的故障模式识别模型,采用特定的征兆群运用加权模糊逻辑进行知识推理诊断出具体故障模式。提出了故障原因、故障影响和故障处理措施在知识库中的查找方法,使得诊断过程更加细致全面。案例分析表明该方法是有效可行的,对汽轮发电机组智能故障诊断系统的设计具有借鉴意义和深入研究的价值。For the low efficiency and poor accuracy of turbo-generator unit＇s fault diagnosis,this paper divides the common 18 kinds of vibration fault into four categories,and takes advantage of support vector machine to distinguish the fault cluster for early fault diagnosis according to the characteristics of vibration signal spectrum.For different fault cluster,different fault pattern recognition model is established.With the use of certain symptom group,this article engages in knowledge reasoning to obtain the specific fault recognition mode by using weighted fuzzy logic.Besides,the searching methods of fault cause,fault influence and troubleshooting measures in the knowledge base are proposed,which make the diagnosis process more meticulous and comprehensive.Case analysis shows that it is feasible to use this method to develop a system for intelligent fault diagnosis of turbo-generator unit,which is valuable for further study in more depth.

关 键 词：支持向量机 知识推理 加权模糊逻辑 故障诊断 汽轮发电机组 

分 类 号：TM621.3[电气工程—电力系统及自动化]