Applying fuzzy cognitive-maps knowledge-representation to failure modes effects analysis

Abstract

A failure mode and effects analysis (FMEA) seeks to determine how a system will behave in the event of a device failure. It involves the integration of several expert tasks to select components for analysis, determine failure modes, predict failure effects, propose corrective actions, etc. During an FMEA, numerical values are often not available or applicable and qualitative thresholds and linguistic terms such as high, slightly high, low, etc., are usually more relevant to the design than numerical expressions. Fuzzy set theory and fuzzy cognitive maps provide a basis for automating much of the reasoning required to carry out an FMEA on a system. They offer a suitable technique to allow symbolic reasoning in the FMEA instead of numerical methods, thus providing human like interpretations of the system model under analysis, and they allow for the integration of multiple expert opinions. This paper describes how fuzzy cognitive maps can be used to describe a system, its missions, failure modes, their causes and effects. The maps can then be evaluated using both numerical and graphical methods to determine the effects of a failure and the consistency of design decisions.