FMEA Group Decision Making And Scoring Correction Method Based On Improved HFPR

With China’s economy shifting from a high growth stage to a high quality development stage,the need for product quality improvement in the domestic environment is becoming more and more urgent,and how to achieve high reliability on the basis of limited cost has become a necessary need.Failure Mode Effects Analysis(FMEA)is an analysis method that analyzes all possible failure modes of each product in a system and all possible effects on the system,and classifies each failure mode and proposes solutions and preventive measures.However,in engineering practice,the traditional FMEA method is not good enough to obtain the real risk ranking of each failure mode of a product because of its single scoring form and simple arithmetic.Considering that FMEA is essentially a multi-attribute group decision making tool,based on the complexity and uncertainty of decision making,this paper introduces the theory of Hesitant Fuzzy Preference Relation(HFPR)for optimizing FMEA research.Although HFPR has good fuzzy and uncertainty properties,the research algorithms for heterogeneous hesitant fuzzy preference information group decision making are not deep enough at this stage,and the existing HFPR processing algorithms are easy to lose the original information of experts,which will lead to the lack of stability and reliability of the results;secondly,the decision results are usually more subjective due to the incomplete information or the contradiction within the group.And the predictability of product failure occurrence degree usually has reference significance for scoring,but the research in such direction is still relatively few at this stage.In order to solve the above problems,this paper researches a hybrid FMEA method and verifies the reasonableness and advancedness of the research method by analyzing the high-voltage pulse output module of a pulse generating device,which is studied as follows.(1)The Multiplicative consistency(MC)theory is extended to handle HFPR with extended MC to improve the original information acquisition of experts in FMEA expert scoring and to extend the processing method to heterogeneous environments.First,a new optimization model is proposed to replace the existing methods based on expectation value or regression method to handle HFPR,and a mathematical model is proposed to obtain both hesitant fuzzy sets and preference set consistency levels by combining the optimization model with the preference set consistency level algorithm,and the feasibility of the model is demonstrated.Next,The model is extended from HFPR to Heterogeneity Hesitant Fuzzy Preference Relation(H-HFPR),and a consistent algorithm model is proposed for Hesitant Fuzzy Linguistic Preference Relation(HFLPR)and Incomplete Hesitant Fuzzy Preference Relation(I-HFPR).Finally,the group consensus theory is used to reach a group consensus in a heterogeneous group decision environment and to determine the risk assessment results of failure modes based on group consensus.(2)A study of the method to revise FMEA scores based on the occurrence predictability principle in product reliability analysis.Firstly,the method of predicting product failure occurrence probability is introduced,and the transformation method of predicted occurrence probability and occurrence degree score is introduced;secondly,the Group Offset Level is proposed based on the expert consensus level and group consensus level obtained from H-HFPR processing;meanwhile,based on the nature of the method of predicting occurrence probability,the concept of caution parameter is proposed to control the correction magnitude of the score;finally,based on the above research,a method based on Based on the above research,we finally propose a method to revise the scoring of each risk assessment parameter of FMEA based on the predicted value of failure occurrence.(3)Implementing the Criticality Analysis(CA)software based on the research method of this paper to analyze the failure of the pulse output module of a pulse generation device and input each failure mode of the module into the CA software for risk assessment to prove the engineering feasibility of the research method of the paper.