Health condition monitoring of Gas Turbine Engine (GTE) components is key for predictive maintenance planning. The task is challenging, as the gas-path components are mostly inaccessible for direct measurements, while at the same time hidden incipient faults must be diagnosed using the available measurements. The presence of multiple faults with similar symptoms adds to the complexity of the diagnostic process. In previous research work, a data-driven multi-mode fault parameter estimation scheme was introduced for real-time multimode diagnosis of GTEs under diverse operating conditions and fault scenarios. In this work, a hybrid diagnostic framework is developed that fuses the results from a measurement-based fault parameter estimation strategy together with a fault propagation model. The hybrid framework uses a novel particle filter (PF) structure with redundant measurements that facilitates updating the particle weights while reducing the dimensionality of the measurement likelihood. Applying the developed framework on GTE gas-path data with four different gradually worsening faults, the results show the diagnostic accuracy increases up to ten times, compared to the previously developed fault parameter estimation scheme.

Additional Metadata
Keywords ANFIS, Fault estimation, Measurement noise, Multi-mode diagnosis, Real-time diagnosis
Persistent URL
Journal Mechanical Systems and Signal Processing
Hanachi, H. (Houman), Liu, J, Kim, I.Y. (Il Yong), & Mechefske, C.K. (Chris K.). (2019). Hybrid sequential fault estimation for multi-mode diagnosis of gas turbine engines. Mechanical Systems and Signal Processing, 115, 255–268. doi:10.1016/j.ymssp.2018.05.054