283 / 2025-06-16 09:00:58

Causal Inference-Based Fault Diagnosis and Abnormal Degradation Detection for Aero-Engine

Fault diagnosis,defect detection,aero-engine

Special Sessions > SSL 06 The reliability of intelligent equipment

Aero-engine fault diagnosis faces challenges such as low accuracy and weak physical interpretability. Additionally, early anomalies are difficult to identify due to complex thermodynamic couplings and nonlinear degradation patterns. To address these challenges, this paper proposes a causal inference-based method. Physically meaningful derived variables are constructed to reflect the status of subsystems, enabling accurate fault diagnosis with machine learning algorithms. Meanwhile, after computing the Conditional Average Treatment Effect (CATE) of variables related to the performance of engine subsystems, abnormal degradation onsets are detected via the Dynamic Sliding-Window Threshold (DSWT) algorithm. Experiments on the N-CMAPSS dataset validate the effectiveness of the proposed method, demonstrating its capability to accurately identify fault modes with high interpretability and enable early identification of abnormal degradation onsets