摘要
典型故障数据样本的严重不足是制约机械故障智能诊断技术发展的主要原因之一。提出了一种基于支持向量机的机械故障诊断新方法,综合了单值和多值支持向量机分类算法,在此基础上,建立了多故障分类器。采用该方法对转子实验台典型的多故障数据进行分类,结果表明:只需少量的时域数据样本来训练分类器,即可实现多故障的识别与诊断,克服了已有方法需要对原始数据进行预处理的困难,可更方便地应用于机械设备多分类故障诊断领域。
The shortage of the typical fault sample is one of the main reasons that restrict the development of intelligent fault diagnose technology. Integrating one-class classification with multiclass classification algorithm, a new mechanical fault diagnosis method based on support vector machine was proposed. Based on the presented algorithm, a multi-fauh classifier was constructed. The multi-fault classifier was applied to analyze to the typical fault data samples obtained by rotor experimental table, the results show that only a small quantity of fault data samples are required in time domain to train the multi-fauh classifier. It can overcome some difficults of the existed method to make preconditioning for orignal vibration signals. Therefore, it can conveniently applied to multi-fault classification of mechanical equipment.
出处
《煤矿机械》
北大核心
2009年第10期243-245,共3页
Coal Mine Machinery
基金
国家自然科学基金资助项目(50805028)
广西自然科学青年基金资助项目(0832082)
广西制造系统与先进制造技术重点实验室主任课题(0842006_023_Z
07109008-012Z)
关键词
支持向量机
多故障分类器
机械故障诊断
support vector machine
multi-fault classifier
machinery fault diagnosis
