Inverter fault diagnosis based on BP neural network

Abstract

Aircraft electric power system (EPS) is critical for the reliable and cost effective performance of aircraft, so the fault diagnosis of the EPS is an important operation and maintenance task. To easily and correctly diagnose any fault indication of DC/AC inverter, some new analysis tools based on the Back Propagation (BP) neural network is being researched by aerospace industry. Inverter is an important DC/AC source of the EPS, some typical fault modes are introduced and the reasons of each fault is analyzed. The BP neural network is used to analyze the faults of the inverter by modelling the faults and training to diagnose the different root causes with typical fault output waveforms. The simulation result verifies this method can diagnose correctly the typical inverter faults, so that it's a practical tool to help the maintenance of the aircraft EPS.