A Comparison Between Two Frameworks for Reliability of System Based on Multi-source Data Fusion

The reliability estimation is difficult for complex system due to the high reliability, huge components and categories of connection structure. Fusing multiple source data collected for system is useful to improve the reliability estimation. In this work, two frameworks about multi-source data integration are presented for reliability estimation based on multiple levels diagram of system. In the first framework, the data in every level are integrated first and transmitted to higher level continually upward system-level for system reliability estimation. In the second framework, multi-source data are grouped into different data types. The data of each data type in lower level are transmitted to higher level separately. If there are data of identical data type in higher level, they are pooled with the transmitted data. By successive data transmission and pooling to system-level, the transmitted data of all the data types are integrated with native data for system reliability estimation. Further, the two frameworks are applied and compared through an illustrative example. The results demonstrate that the second framework is more robust than first framework, due to the fact that data integration is required only once in system-level for the second framework while it is necessary in each level for the first framework.