Abstract: There is a large amount of uncertainty in the development and operation of satellites, which directly affects the satellite performance and lifetime. How to realize the uncertainty modeling is an urgent problem, and the multi-source information integration is the key to uncertainty modeling. In addition, it is difficult to obtain the distribution information of the system directly due to the constraints of time and cost. To realize the multi-source information integration of satellite systems with a lack of information, the methods of the multilevel system structure modulization and the system information synthesization were put forward based on the traditional Bayesian melding method. Through the system structure decomposition, the information of lower levels was merged into the higher levels. Then with the induction inference method, the system node could get new information instead of non-information. Through a case study of a satellite attitude control system, the proposed method can make full use of the system structure relationship, effectively integrate and update the distribution information of each component and subsystem. The problem of missing system information is solved, which will provide reference for further development of the satellite system uncertainty design and optimization.

Key words: satellite system, uncertainties, uncertainty modeling, Bayesian melding method, multi-level structure, multi-source information