Reliability analysis of station specific equipment unit in field stations based on improved D-S evidence theory
Author of the article:BAO Mingyu1, CUI Mingfang1, PENG Xingyu2, LIU Chang1
Author's Workplace：1. Safety, Environment & Technology Supervision Research Institute, PetroChina Southwest Oil & Gasfield Company, Chengdu, Sichuan, 610041, China; 2. Petroleum Engineering School, Southwest Petroleum University, Chengdu, Sichuan, 610500, China
Key Words：Expert scoring method; Reliability analysis; D-S evidence theory; Evaluation rating; Confidence level
Abstract：When the expert scoring method was applied on reliability analysis of specific equipment unit in a field station, there were differences in the evaluation of the attribute of the same thing due to constraints and limitations of past experience and personal preferences, and the decision of the final evaluation rating would affect the calculated results of the failure probability. Aiming to resolve the problem of decision-making whereby the attribute evaluation result was a linguistic variable and the expert evaluation ratings were inconsistent, a decision-making method based on improved D-S(Dempster-Shafer) evidence theory was put forward for evaluating the result of reliability analysis. This paper introduced the principle of D-S evidence theory and the conventional evidence conflict processing method. Considering the fact that the traditional evidence theory does not consider the confidence level of the evidence source, this paper introduced the coefficient of expert authority level and the expert evaluation consistency coefficient, which proposed that the confidence level of experts is to be determined as the confidence level of evidence source. The credibility of the evidence is then corrected accordingly, and the final expert evaluation rating can be determined and applied to the reliability analysis of the equipment. The calculation from a case study verifies that the method can effectively arrive at the decision of the final evaluation rating and make the reliability analysis results more accurate.