Conversion of Reliability Data Based on Optimization Algorithm
In the reliability engineering, many methods for synthesis of system reliability are based on binomial (success/failure) data. When several different types of component reliability data are contained in a complex system, the evaluation of the system reliability becomes difficult. In order to get the reliability of such system, non-binomial data is commonly converted to binomial data by the estimate and variance method, the two point method, the estimate and lower confidence limit method or the moment method. This paper presents a new method to convert non-binomial reliability data to binomial data. Beta distribution which describes the reliability confidence distribution of binomial data is used to approximate the reliability confidence distributions of the non-binomial data by the method. Two points on the original confidence distribution are select as the control points. When the conversion error is chose as the objective function, the two best control points are found by the optimization algorithm. By the two best control points, the best beta distribution with lowest conversion error is obtained, so is the equivalent binomial data. An example is given. The result shows that the method proposed in this paper is valid and the precision is the highest.
reliability confidence distribution data conversion optimization non-binomial data
WANG Xue FENG Zhigang GAO Puyun
Department of Astronautical Science and Engineering, School of Aerospace and Materials Engineering,National University of Defense Technology, Changsha, Hunan, P.R.China, 410073
国际会议
威海
英文
635-641
2010-07-24（万方平台首次上网日期，不代表论文的发表时间）