Failure mode and effects analysis based on a novel fuzzy evidential method

A comprehensive way to rank the risk of failure modes is proposed by fusing the feature information of O, S and D with Dempster–Shafer (D–S) evidence theory and the experimental results show that this method is reasonable and effective for real applications.

Abstract

Failure mode and effect analysis (FMEA) has been widely applied to examine potential failures in systems, designs, and products. The risk priority number (RPN) is the key criteria to determine the risk priorities of the failure modes. Traditionally, the determination of RPN is based on the risk factors like occurrence (O), severity (S) and detection (D), which require to be precisely evaluated. However, this method has many irrationalities and needs to be improved for more applications. To overcome the shortcomings of the traditional FMEA and better model and process uncertainties, we propose a FMEA model based on a novel fuzzy evidential method. The risks of the risk factors are evaluated by fuzzy membership degree. As a result, a comprehensive way to rank the risk of failure modes is proposed by fusing the feature information of O, S and D with Dempster–Shafer (D–S) evidence theory. The advantages of the proposed method are that it can not only cover the diversity and uncertainty of the risk assessment, but also improve the reliability of the RPN by data fusion. To validate the proposed method, a case study of a micro-electro-mechanical system (MEMS) is performed. The experimental results show that this method is reasonable and effective for real applications.