加权证据理论信息融合方法在故障诊断中的应用  

APPLICATION OF WEIGHTED EVIDENTIAL THEORY AND ITS INFORMATION FUSION METHOD IN FAULT DIAGNOSIS

作　　者：谭青[1] 向阳辉[1] 

机构地区：[1]中南大学机电工程学院,长沙410083

出　　处：《振动与冲击》2008年第4期112-116,共5页Journal of Vibration and Shock

摘　　要：为了综合合理利用设备多个方面的故障特征信息,提高故障诊断的准确性,提出了一种多故障特征信息融合的故障诊断方法。该方法依据加权改进后的证据理论,通过构建融合诊断框架,分析融合组建原则,保证了各特征域故障诊断过程中存在的不确定性经过融合后能够最大限度相互削弱,从而从理论上降低了故障诊断的不确定性。转子实验结果表明,多故障特征信息融合后的诊断结果可信度明显提高,不确定性显著减小,充分验证了该融合诊断方法能够有效提高故障诊断的准确性和可靠性。A fault diagnosis method of multi-fault characteristic information fusion is proposed to improve accuracy of fault diagnosis using integrated multi-fault characteristic information.The method is based on the evidence theory improved by the weighted concept.By constructing the fusion diagnosis frame and analyzing the fusion diagnosis principle,uncertainties in all characteristics of fault diagnosis can be weakened each other to the utmost.So the uncertainty of fault diagnosis can be reduced in theory.The result of a rotor experiment shows that the reliability of fault diagnosis upon the multi-fault characteristic information fusion is improved evidently and its uncertainty decreases remarkably.It proves that the proposed method can improve the accuracy and reliability of fault diagnosis.

关 键 词：证据理论 信息融合 故障诊断 概率分配 不确定性 

分 类 号：TP306[自动化与计算机技术—计算机系统结构]