System reliability analysis for independent and nonidentical components based on survival signature

Abstract

Survival signatures have been widely used for analyzing arbitrary system structures having independent failure events. It is challenging to study system reliability when diverse components are present since the fundamental premise of the survival signatures is that components should be interchangeable within the same subsystem. In this research, two methods based on the fundamental survival signature idea are suggested to examine the reliability of complex systems with independently but not necessarily identically distributed (INID) components. The first algorithm is based on the weighted random sampling (WRS) method to calculate the survival signatures. The second algorithm adopts the idea of divide-and-conquer in computing the probability structure. The application of these two algorithms and the exhaustive algorithm in analyzing the example systems are presented. The results show that the two algorithms can significantly reduce the computation time compared to the traditional methods. Finally, the developed algorithms are used in the reliability analysis of a real practical problem, the Tsinghua campus water supply pipeline system. The results regarding the predictions of system reliability as well as the component importance index are discussed.