基于Dempster-Shafer证据理论的信息融合在设备故障诊断中应用

Research on fault diagnosis method based on Dempster-Shafer evidential theory

给出了一种基于神经网络的基本概率分配构造方法和诊断决策规则,提出了一种基于Dempster-Shafer证据理论的多故障特征信息融合的故障诊断方法,并以旋转机械故障诊断为例,详细说明了该方法的具体实现步骤.结果表明,经过多故障特征信息融合,诊断结论的可信度明显提高,不确定性明显减小,因此充分显示了该诊断方法的有效性.

A method based on artificial neural network of basic probability assignment and diagnosis decisionmaking rules is proposed. A new approach to fault diagnosis based on multiinformation fusion of fault characteristic and DempsterShafer evidential theory is put forward. Taking the rotary machinery fault diagnosis for example, the implementing process of this method is elaborated in detail. The results indicate that the reliability of the diagnostic result is improved evidently and the uncertainty decreases markedly. So the fault diagnosis method proves highly effective.