关键词: 动态主元分析, 证据理论, 故障诊断, 多信息融合

Abstract: In order to comprehensively and reasonably utilize multi-source information of equipments to improve the accuracy of fault diagnosis, a method of fusion fault diagnosis is proposed based on Dynamic Principle Component Analysis（DPCA） and improved evidence theory. This method constitutes multi evidences to fault diagnosis at many levels by means of DPCA and revises the basic assignment probability according to the authoritative coefficients based on statistical errors. The method of time authoritative conversion of evidence and conflict weighted assignment are proposed to improve evidence combination rules. The experimental results show that the weighted fusion treatment of multi-source information evidences can reduce the conflicts based on single information, which can increase reliability by 50% and greatly reduce uncertainty. The results also are unaffected by the decline of evidence authority. So the method can effectively improve the accuracy of fault diagnosis.

Key words: dynamic principle component analysis, evidence theory, fault diagnosis, multi-source information fusion