Research on fault pattern recognition for aircraft fuel system with its performance simulation

This paper presents application research on fault pattern recognition for aircraft fuel system based on its performance simulation. Fault pattern recognition method is used to perform fault detection, fault isolation, fault prognostics and so on in complicated system, which is basic research contents of prognostics and health management (PHM). Since the hardware of fuel system for an aircraft is too expensive in laboratory, it is better to build simulation software to research on the method of fault pattern recognition for fuel system of an aircraft. The simulation software of fault pattern recognition is able to carry out performance simulation for the aircraft fuel flow parameters such as pressure variation, flow rate in the fuel system pipe network both on the fuel system normal condition and its malfunction condition. When a fault pattern of a component such as radial pump, ball valve happens, the nodes in which maximum pressure variation has happen in the fuel system pipe network are selected, and the virtual sensor is arranged in these nodes. Through the signal correlation comparison of the virtual pressure sensors under the normal condition and malfunction condition, the fault pattern is recognized. Since the fault pattern recognition methods is at its very initiative stage, the false alarm rate is not low. Nevertheless, the simulation software is able to verify and study the new concept fault pattern recognition methods, which will supply technical support for the onboard aircraft PHM system in the future.