Equipment health management through information fusion for reliability

A generic framework for estimating the reliability of equipment is through “information fusion” of its failure history, predictive maintenance data and domain expert´s knowledge is proposed and demonstrated. The framework uses “Degree of Certainty” arrived at using fuzzy sets and “Belief” & “Plausibility” measures to arrive at a decision on the effectiveness of predictive maintenance. Uncertainty, randomness, imprecision and ambiguity/conflicts inherent in the information/data are logically synthesized. The information/data is grouped into information blocks and the framework connects these blocks, synthesizing the information to arrive at logical conclusions. The essence of the proposed framework is to provide information on missing links between quantitative data and qualitative expert domain knowledge, which are ignored while dealing with reliability metrics such as MTBF, MTTR etc. This framework is demonstrated with a case study using the maintenance information/data of large industrial motors. The case study highlights the advantages of the framework used to fuse the ma intenance information/data drawn from various sources to draw inferences on the failure processes of the motors. This in turn leads to a revision of the maintenance strategy of the motors and can lead to large tangible/intangible benefits in reducing the failures. A simple desktop application program can be developed based on this framework to suit individual plant operations.