Synchronising Design Knowledge and Real-Time Monitoring Information for Aircraft Complex System Diagnosis

With increasing complexity of aircraft systems and various operational environments, fault modes and effects among different hierarchical system levels can be hardly described by accurate physical models. This brings uncertainty in fault location and maintenance. Motivated by the low effiency in trouble-shooting for aircraft complex systems, this study proposes an intelligent diagnostic approach that can synchronize design knowledge and real-time monitoring messages to make full use of available information. An aircraft brake system is selected as a typical electromechanical system with frequent fault occurrence. A multi-level Bayesian network is created based on fault mode and effect analysis. Conjugate distribution is employed to combine prior knowledge and observation for statistical inference update. The proposed method can reduce the diagnostic uncertainty and provide more effective trouble shooting.