Diagnostic system configuration optimization and its application to aircraft engine fault diagnosis

Conventional fault diagnostic system design is typically performance-driven, which has several drawbacks. The most significant one is that the performance-driven design may result in an increase on overall maintenance costs. To address these problems, this thesis introduces a novel design philosophy for fault diagnostic systems. Under this new design philosophy, a fault diagnostic system design is casting as an optimization problem where the overall maintenance cost is the objective function and the diagnostic system configurations are the optimization variables.; The new design method consists of a cost model and optimization methods, among other critical components. The former allows for calculating the cost/benefit for a potential FDS design configuration, while the latter is the solver for finding the optimum solution. Realizing that solving the FDS configuration optimization is computationally prohibitively expensive, which may not be practical for typical real world problems, the thesis proposes a distributed optimization scheme that decomposes the FDS configuration optimization into a series of small-scaled sub-problems that can be solved independently in parallel. For validating and demonstrating the design details of the new design method, a real-world aircraft engine fault diagnostic system is designed using the proposed new design method.; Aircraft engines are operated at various points in flight regime (different altitudes and Mach numbers), which cause engine performance parameter changes. The flight regime induced engine performance parameter changes disturb the parameter patterns associated with engine faults, thus make the engine fault diagnosis notoriously difficult. This thesis tackles the flight regime issue using an innovative method, namely flight regime mapping. The proposed flight regime mapping essentially compensates for flight regime induced parameter changes, thus accentuates the engine condition related changes, by mapping the engine parameter values from actual flight regime to sea level static. The flight regime mapping results in an improved performance of AEFD systems.