Complex System Maintainability Verification under Mixture Distribution

Abstract

Maintainability verification is an important work in system development. However, two problems arise when the maintenance time of complex system conforms to a mixture distribution. The first problem is the occurrence of assuming log-normal distribution instead of mixture distribution and the second one is the limited sample size for maintainability verification. Regarding the problems mentioned above, firstly, the maintenance sample generation process of complex system is analyzed and the mixture distribution model of maintenance time is proposed. Then, the changes in two types of test risks are analyzed when assuming log-normal distribution instead of mixture distribution. Finally, a method to perform maintainability verification using similar maintenance task’s samples is proposed. A case study using the maintenance time data of satellite communication interface system shows the efficiency of the proposed method.