IoT-Enhanced Fault Diagnosis and Two-Stage RUL Prediction Method for Aircraft Hydraulic Systems Based on Sensor Data

Fault diagnosis and remaining useful life (RUL) prediction technologies are critical for aircraft hydraulic systems to ensure flight safety and optimize maintenance schedules. This article presents an Internet of Things (IoT)-enhanced fault diagnosis and two-stage RUL prediction method for aircraft hydraulic systems based on sensor data. Initially, a three-layer IoT monitoring framework is constructed to enable real-time acquisition of critical hydraulic system parameters through multiple heterogeneous sensors. Subsequently, failure mode and effects analysis (FMEA) and fault tree analysis (FTA) are integrated to identify failure modes and analyze fault propagation pathways. Following this, an improved kernel principal component analysis (Kernel-PCA) method is developed to achieve dimensionality reduction while preserving critical degradation features in high-dimensional sensor data. A two-stage RUL prediction model is then proposed. The first stage employs seasonal and trend decomposition using loess (STL) combined with multiobjective optimized ensemble learning to predict trend and seasonal components. The second stage utilizes a 1-D CNN-BiLSTM architecture to capture residual patterns. These prediction results are subsequently integrated to generate accurate RUL estimates. The primary advantage of this work lies in the proposed integrated framework’s capability to effectively address complex fault characteristics and high-dimensional sensor data processing challenges while enhancing RUL prediction accuracy through the two-stage prediction model. The experimental results on domestically manufactured large aircraft hydraulic systems demonstrate that the proposed method significantly outperforms traditional approaches, achieving root-mean-square error (RMSE) and mean absolute error (MAE) values of 21.00 and 15.34, respectively.