Sensitivity Analysis For Random Evaluation Parameters Of Process Failure Risk Assessment

Risk priority number abbreviated as RPN is applied to assessment for traditional process failure mode in the event of possible risks. It is determined by the product of severity abbreviated as S, occurrence abbreviated as O and detection abbreviated as D.However, it is applicable to mass customization production mode. In order to solve the problem for analysis of process failure mode of small batch customization production,the author’s research team put forward the multidimensional analysis model of risk factor, fuzzy confidence interval evaluation model, and RPN evaluation method based on the generalized Hausdorff distance. For the purpose of accurately acquiring the influence degree of random evaluation parameters of risk assessment of fuzzy confidence interval evaluation model, it is necessary to carry out sensitivity analysis for random evaluation parameter. Through sensitivity analysis, multidimensional analysis model of risk factors can be optimized.Paper first expounds fuzzy confidence interval assessment model for process failure risk assessment model and RPN evaluation model based on the generalized Hausdorff distance. Each parameter’s characteristics of fuzzy confidence interval evaluation model of risk factors was analyzed and researched through the two models.S,O and D were evaluated in the form of ? ba,? zy,.Interval parameters,value orientation and confidence parameters of each risk factor value are random parameters.The characteristics of each parameter is not only related to the order determined by the valuer but also related to the parameter values determined in advance. That is to say,there’s coupling relationship among random evaluation parameters.What’s more,according to the determined sequence of a, b, y and z, the probability distribution model of each parameter was determined one by one. Specific process is as follows.First, the evaluation system of parameter a was constructed. Second, the probability distribution of a was determined based on the maximum entropy principle. Third, the general probability distribution model reflecting the characteristics of each parameter was built on the basis of coupling relationship and the value of each parameter.Eventually, probability distribution of each parameter was achieved quantitatively by fitting of undetermined coefficients.The sensitivity sequence of each random evaluationparameter was acquired based on the RPN evaluation method of generalized Hausdorff distance and Monte Carlo statistical simulation method. And the expectation of each random evaluation parameter was approximately obtained by law of large Numbers.According to the expectation,the sensitivity of each parameter was sorted. After the above work, the research results were applied to sensitivity analysis for parameters of process failure risk assessment of fuzzy confidence interval. Application study showed that the highest sensitivity parameter was confidence and the lowest was deviation tendency. So the evaluation criteria of confidence should be set reasonably and accurately. Meanwhile, it is necessary to consider reducing level classification of deviation tendency for attenuating experts thought burden and improving the work efficiency.Research work for this paper is a helpful attempt for carrying out systematically sensitivity analysis of each random parameter of risk assessment model of fuzzy confidence interval process failure. The above research results were provided with good reference value for Multidimensional evaluation model of risk factors and optimization of evaluation criteria.