Integrating a consensus-reaching mechanism with bounded confidences into failure mode and effect analysis under incomplete context

Failure mode and effect analysis (FMEA) is one of the most powerful pre-accident risk management techniques and has been extensively used in various fields. Nevertheless, consensus issues and incomplete assessment information are rarely investigated in the current FMEA method. This paper integrates a consensus-reaching mechanism into the FMEA framework and adopts incomplete linguistic distribution assessments to represent risk assessment information. In the constructed FMEA framework, a deviation minimum-based optimization model is presented to manage incomplete assessment information, and this model aims to minimize the opinion deviation among FMEA members. Then, a consensus-reaching model with a novel feedback adjustment mechanism is proposed to assist FMEA members to gain a consensus. The novel feedback adjustment mechanism is based on a natural assumption that individuals have bounded confidences when updating opinions and that they only take into account the adjustment suggestions that are similar to their own. A two-stage optimization model with bounded confidences is built to support the novel feedback adjustment mechanism. In the first stage, a consensus maximum-based optimization model is proposed to judge whether the consensus among the FMEA members can be achieved under the specific bounded confidences. If it can be achieved, then in the second stage, an adjustment minimum-based optimization model with bounded confidences is designed to generate feedback adjustment suggestions: otherwise, the feedback adjustment suggestions generated from the first stage are used for supporting modifying opinions. Two case studies show that the proposal has a good application value, and the simulation experiment shows that our proposal can effectively improve the consensus success ratio and accelerate the speed to achieve a consensus compared with the existing baseline approach. (C) 2019 Elsevier B.V. All rights reserved.