基于条件证据的信息融合故障诊断方法

该文基于观测信息与先验信息和谐的思想,结合条件证据理论,提出了一种融合先验信息的故障诊断方法。首先将传感器获取的关于设备运行状态的观测数据,通过模糊信息的随机集表示方法转化为D-S证据的随机集形式;然后计算所得证据与先验信息之间的和谐度,最后利用条件证据理论将需要融合的证据进行组合,从而做出诊断决策。该方法可在复杂运行环境下充分利用各种信息,提高故障诊断的可靠性。

Fault Diagnosis Method Based on Conditioned Dempster-Shafer Theory

Based on the concept of the concordance existing between the measurement evidences and the prior knowledge, and on the conditioning Dempster- Sharer evidence theory, this paper provides a novel fault diagnosis approach which reflects the influences of prior knowledge. First, data about equipment＇ s running information acquired from sensors are transformed to the form of bodies of D - S evidence based on random set model of fuzzy in- formation. And then, we apply the conditioning D - S evidence theory to combine these evidences, and calculate the concordance. This method can help us to increase the reliability of fault diagnosis under complex running circumstances.