Fault Diagnosis of Aircraft Power System Based on Long Short-Term Memory Network

Abstract

The aircraft power system is the source of electrical energy for all onboard electrical equipment, and plays an important and irreplaceable role in the aircraft. Aiming at the problem that traditional aircraft power system fault diagnosis methods often require the establishment of accurate mathematical models and cannot quickly perform fault diagnosis and identification, this paper proposes an aircraft power system fault diagnosis method based on long and short-term memory networks. Taking the typical components of the aircraft power system integrally driving the generator as an example, the fault diagnosis research is carried out. The simulation model is used to collect fault data to make a fault simulation data set. After many parameter tuning experiments, an ideal fault diagnosis model is obtained. Under the existing experimental conditions and data sets, the fault diagnosis accuracy rate reaches 98.57%, which can meet the needs of actual work to a certain extent.