A TFPG-Based Method of Fault Modeling and Diagnosis for IMA Systems

Abstract

Timed Failure Propagation Graph (TFPG) is an effective method to modeling and analyzing failure/fault conditions of complex dynamic systems such as fuel transfer systems, GPS assembly and ADIRU assembly. This paper proposes an approach to characterize the system behavior under failure/fault conditions of integrated modular avionics (IMA). According to failure mode and effects analysis (FMEA) of related line replaceable units (LRU) and modules, fault tree analysis and real-time requirements, the TFPG-specific meta-model and domain models of IMA are presented by the Generic Modeling Environment (GME) for the purpose of fault diagnosis. Based on the GME, graphical modeling and other functions are realized by building domain-specific meta-model. The process of generating XML files from a specific model is implemented, which enhances the universality of fault model in different platforms. The fault diagnosis algorithm based on TFPG is studied. Using the algorithm of backward inference, the fault candidates are listed on the basis of logical and temporal consistency as well as measurements of plausibility. Considering the characteristics of the system and the types of avionics faults, the directed graph model of a functional failure of IMA is constructed, and the process of establishing the TFPG model based on the GME is given. The result shows the effectiveness of TFPG-based diagnosis in FDE relating and cascading effects removal.