Abstract: To meet the requirements of completeness, accuracy and real-time performance for the on-board fault diagnosis system of a spacecraft, this article proposes a novel qualitative diagnosis approach based on hierarchical digraphs. The hierarchical strategy for digraphs reduces search space for failure sources; an inference algorithm is proposed to identify fault propagation consistent branches by making use of the qualitative relations between the measured nodes, and to search backward the set of failure source candidates. Furthermore, the failure source candidates are ranked according to the probability rate defined by the product value of the failure propagation weight and the failure rate of the failure source candidate. The diagnosis system for the primary power system of a certain satellite is modeled with the approach, and the simulation test for the system shows that the approach is highly efficient, and the diagnosis result is complete and accurate. This suggests that the approach may be suitable for application in spacecraft on-board diagnosis systems.

Key words: diagnosis, digraph, fault propagation path, consistent branch, the primary power system