| 摘要 |
Methods and apparatus for identifying faults in a vehicle, such as an aircraft (2, figure 1), that includes a plurality of vehicle subsystems (4-12 figure 1), the method comprising: for each vehicle subsystem (4-12 figure 1) receiving and processing information from that subsystem, using a respective subsystem reasoner 30-38 corresponding to that subsystem, to determine probability for faults in that subsystem; receiving and processing information from at least two different sources (selected from the vehicle subsystems (4-12 figure 1) and the subsystem reasoners 30-38), using a vehicle health reasoner 40, to determine, for each of one or more further faults, a probability that that further fault has occurred, each further faults being a fault the symptoms of which manifest in more than one of the subsystems; and displaying the identity of the most likely fault. The method may identify tests which can further refine its level of confidence if performed. The method may method such as probabilistic Bayesian networks, neural networks, support vector based processes, regression based processes, decision tree based processes or fuzzy logic based diagnostic processes to determine the probabilities of faults. |
