Spacecraft Fault Diagnos Is Based On The Fault Tree

Fault diagnosis technology in the field of aerospace has achieved a great success with the development of computer technology and artificial intelligence technology. However, these diagnostic systems are poor performance in adaptive capacity, learning ability, knowledge acquisition, and processing imprecise information, because the majority of the current domestic spacecraft fault diagnosis system is still using a single diagnostic methods. Therefore, how to integrate a variety of smart technology and study a hybrid intelligent diagnostic model, that has an important practical significance to promote an intelligent and scientific development in spacecraft fault diagnosis.Based on the system research about the application technology of spacecraft fault diagnosis, combining with the actual features of spacecraft monitoring management, the following works has been done:1. In order to meet the demand on the status monitoring, the fault location and the early prediction during the process of spacecraft monitoring and controlling, hybrid Intelligent integration mode based on the technology of RBP (Rule Based Reasoning) and FT (Rule Based Reasoning) was adopted, and the system framework and work flow of spacecraft hybrid intelligent diagnosis was designed. This method can be identified by way of online access to relevant signs of the credibility of the fault, the fault tree logic structure using the tight correlation characteristics of reasoning;2. In order to overcome the inherent defect of the binary logic-based comparison method, according to the idea that fault signs were classified by the condition of spacecraft, I proposed to use the numerical characteristics of telemetry data to screen the telemetry outliers, and I have found a new inference rule about uncertainty measure of fault symptoms. This method can overcome the shortcomings of binary threshold whose boundaries is not easy to determine, the data state is abrupt, the use of data message is not sufficient the in the criteria process;3. Fault tree's node-type knowledge representation and reasoning mechanisms are used to handle the spacecraft's procedural knowledge and declarative knowledge. Two-way mixed-reasoning fault tree was designed to realize the incentive location and the early prediction of spacecraft fault. The feature of this method is that the positive reasoning is used for fault prediction, while backward reasoning is used for implementing fault location function. 4. SESRF (Spacecraft Expert System Combining RBR and FT) system was designed, implemented, and used to test an actual fault model; the test results verify the system is feasible.