Abstract
Failure mode and effects analysis (FMEA) is a typical risk assessment and prevention technology, in which different experts provide different assessments on the target system to identify the risk grades of its components. Sometimes, the assessments contain conflict information. How to manage and fuse the conflict assessment information is an open issue. We propose a triangular distribution-based basic probability assignment (TDBPA) method to model and fuse the conflict risk level coming from different experts' assessments in the frame of Dempster-Shafer evidence theory. First, the subjective assessments of risk analysis from domain experts are modeled with belief structure in Dempster-Shafer evidence theory. Then, the assessments are transformed as the TDBPA function. Thirdly, the conflict risk assessments from the FMEA team for failure analysis can be fused with Dempster rule of combination. After that, the modified risk priority number (RPN) model based on fused assessment can be calculated for ranking of failure modes. Finally, recommended actions should be taken for prevention of potential risk items. We verify the rationality and efficiency of the proposed method with a case study in the blades of an aircraft turbine. In short, the presented FMEA methodology procedure in this paper is well organized so that we can apply it in a more simple and understandable way. Utilizing the character of triangular distribution, taking adjacent values into account, TDBPA method can smooth the conflict assessment for information fusion. In addition, the shortcoming of repeating values in classical RPN is eliminated in the proposed method, which improves the ability for risk assessment of FMEA.
Highlights
Since it was introduced by NASA in 1960s [1], failure mode and effects analysis (FMEA), as an analytical tool in reliability, has been proved to be remarkably effective and applied extensively in many fields such as risk evaluation [2], decision making [3]–[5] and so on
The triangular distribution-based basic probability assignment (BPA) (TDBPA) method can smooth the conflict assessment for information fusion by taking into consideration of the adjacent value of a risk level assessed by an expert
The improved Failure mode and effects analysis (FMEA) incorporating TDBPA method is validated to be efficient in the blades of an aircraft turbine
Summary
Since it was introduced by NASA in 1960s [1], failure mode and effects analysis (FMEA), as an analytical tool in reliability, has been proved to be remarkably effective and applied extensively in many fields such as risk evaluation [2], decision making [3]–[5] and so on. To inherit the advantage of fuzzy sets theory as well as the precise model and fuse the subjective assessments from experts, we propose an improved and simple method for generation of BPA. With the triangular distribution-based BPA (TDBPA) method, the conflict integer values in risk level coming from subjective assessments of FMEA experts can be modeled and fused for failure analysis and prevention. Note that, when experts assign different but precise values to risk factors, the basic probability assignment (BPA) constructed becomes highly conflicting evidence, which cannot be fused by Dempster combination rule and will result in errors in FMEA. The aforementioned methods do not put emphasis on reconstructing the BPAs. To improve the modeling of subjective assessment in [20] by generating BPA more precisely, the work in [13] introduces the Gaussian distribution function, while which seems to be more complex in practical engineering. Sometimes, the K is defined as a conflict coefficient between two BOEs
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
