A Twice-Corrected Method for Leakage Failure Prediction and Performance Evaluation of Hydraulic System

Hydraulic systems are used in a wide range of aerospace, mechanical and other applications, but problems such as leaks and fluctuating oil levels often lead to failures. Existing fault diagnosis methods ignore the gradual surge in hydraulic oil quantity during shutdown and the hydraulic oil surge in hydraulic oil quantity when the brake is released, resulting in inaccurate monitoring and unreliable assessment. To address this problem, we propose a twice-corrected method for leakage fault prediction and performance assessment of hydraulic systems: an extrapolation correction algorithm and a brake release correction algorithm are introduced to mitigate the fluctuation of hydraulic oil quantity during hydraulic return and brake release, thus improving the accuracy of the data. Subsequently, a history threshold model was constructed using Bayesian inference to provide a theoretical basis for fault diagnosis. Finally, the hydraulic oil quantity can be predicted dynamically using the autoencoder long short-term memory (AE-LSTM) model, which detects faults in advance and optimizes the maintenance strategy, thus improving the stability and efficiency of the system. Experimental results show that our method can accurately predict the trend of hydraulic oil quantity, and can effectively perform fault diagnosis and performance evaluation.