以组合故障案例为证据的一种诊断推理方法

Diagnosis Reasoning Method Based on Evidence with Case Contained Multiple Fault Modes

针对单纯依靠案例特征之间的相似性进行推理、结果中将存在不确定性信息的问题,通过引入证据理论,将搜索结果中的故障模式构成目标识别框架,并基于案例之间的相似度,给定各故障模式的基本概率赋值.利用证据理论的组合规则,对搜索出的相关案例样本进行信息融合,从全局相似度中有效地分离出案例对特定故障模式的确定性信息,从而降低了推理结果中故障模式的不确定性信息.对大机组实际故障数据的应用,证明了方法的有效性,为组合故障的案例推理提供了解决思路.

A complex relationship between case features and fault mode in multi-fault mode case usually leads to reasoning results containing uncertain information, which motivates to propose a method based on D-S evidence theory. Via constructing the frame of discernment composed of fault modes and giving basic probability assignment based on similarities among cases, the method based on D-S evidence theory enables to extract certain information of special fault mode from global similarity among cases. The examples verifies the higher validity of this method for reducing the uncertain information contained in reasoning result than the conventional methods.