Multiple Model-Based Hybrid Kalman Filter for Fault Diagnosis of Jet Engines

This paper studies a multi-model-based hybrid Kalman filter fault diagnosis method. This method uses a non-linear airborne model to adapt to the level of engine degradation, and uses probability density functions to perform hypothesis testing. It can effectively solve the problem of threshold selection and evaluation. The paper establishes the working state model of aero engine. By analyzing various fault situations of sensors, components and actuators, fault-free filters and filter banks of sensors, components and actuators are constructed. By using a Gaussian density function recursive algorithm to locate the fault, the credibility ranking of the fault occurrence site is generated to achieve the purpose of fault isolation. Simulation results show that the algorithm can achieve effective detection and isolation of aero-engine gas path faults, and meet the real-time requirements of airborne computers.