A New FMEA Approach with a Consensus Reaching Model Considering Bounded Confidences Based on Deep Learning Classification Algorithm

Failure mode and effects analysis (FMEA) is a traditional methodology applied to identify, classify, and resolve system failures. In order to optimize this method, we proposed a new FMEA model by combining the consensus model with optimization feedback mechanism and the K-means algorithm. Firstly, the consensus model with optimization feedback mechanism is used to enhance consensus and mutual acceptance among experts within their bounded confidence. It is designed to maximize the similarity of the team, effectively promoting collaboration and agreement among experts. By considering experts' bounded confidence, the model measures the level of consensus in terms of expert acceptance. Next is the application of the K-means algorithm. The algorithm is used to cluster twelve faults into three categories: "most critical", "general critical" and "least critical". According to the clustering results, the "most critical" failure modes are prioritized for correction. Finally, an endoscopic risk assessment exercise will be used to show the feasibility and benefits of the new FMEA method.