A fully data-driven FMEA framework for risk assessment on manufacturing processes using a hybrid approach

Abstract

The Failure Mode and Effect Analysis (FMEA) is a widely used method that effectively identifies and prioritizes potential risks in a given system or process. However, traditional and modified versions of FMEA are often criticized for their subjective assessments, inadequate risk prioritization methods, and lack of consideration of the importance level of risk factors. To address these issues, this study introduces a data-driven FMEA approach. Specifically, the proposed approach utilizes data-driven risk factors to determine objective rankings of failure modes. This study uses the frequency and stability of failures, time and product loss cost due to failure as objective and data-driven risk factors. These factors enable a more precise description of the influence of risk factors on failure modes. The Modified Criteria Ranking Importance with Intra-criteria Correlation (M-CRITIC) method is employed to assign weights to the identified risk factors, which indicates their level of importance in analysis. Additionally, the recently proposed Alternative by Alternative Comparison (ABAC) method is used to derive the risk priorities of failure modes. The effectiveness and applicability of the developed approach are demonstrated through a case study focused on manufacturing process risk analysis in the food industry. Furthermore, this study contributes to the growing trend toward objective risk calculations for FMEA and highlights the importance of using data-driven models for risk analysis.