A Novel Method Based on Fused Physical Model and Adaptive Threshold for Landing Gear Buffer Strut Fault Detection and Prediction

Landing gear reliability is critical to aviation safety and operational efficiency. Traditional methods, such as periodic inspections and fixed-threshold fault detection, cannot perform real-time performance evaluation and early fault detection. In this paper, we propose a landing gear fault detection and fault prediction method based on fused physical model (FPM). By fusing the landing gear fault physical model, adaptive threshold and the data-driven method, we compare the fault physical model with the real operation data to quantify the landing gear operation state and input it into the prediction network as features. The prediction network adopts AE-LSTM (proposed), which can reduce the prediction error and provide more robust fault prediction results than traditional RNN and ARIMA, and enhance the fault prediction and health management capability of aerospace system equipment.