基于最小决策风险的故障诊断方法

Fault Diagnosis Method Based on Minimal Decision Risk

中文关键词:  预测和健康管理  故障诊断  支持向量机  最小决策风险

英文关键词:PHM  fault diagnosis  support vector machine  minimal decision risk

中文摘要:

      预测与健康管理技术对故障诊断的准确性提出了更高的要求。在传统基于支持向量机的故障诊断方法中引入最小决策风险，即将先验故障模式信息与数据驱动学习算法相融合，以获得更为有效的故障诊断结果。给出了基于多分类后验概率最小决策风险的SVM故障模式识别的实验步骤，并选取某电路板的400组数据进行实验。结果表明，提出的故障诊断方法可有效减少故障的漏报率，提升系统整体的诊断准确性。

英文摘要:

      Prognostic and health management (PHM) sets higher demands for the accuracy of fault diagnosis.The minimal decision risk in traditional fault diagnosis methods based on support vector machine(SVM)is brought,which combines priori fault mode information and data-driven learning algorithm for more effective fault diagnosis results.A procedure for SVM fault diagnosis based on minimal decision risk of posterior probability in multi-classifier is proposed,and an experiment is done with 400 sets of data of a circuit board.Experimental results show that the method presented can reduce the failure rate of missing report effectively and enhance the diagnostic accuracy of the system.