Research on fault symptom monitoring and early on fault symptom monitoring and early warning method of typical transport system

To enhance the reliability and safety of aircraft flight testing, this study introduces an interpretable multi-parameter fusion methodology designed for monitoring and early warning of fault symptoms in typical transportation systems. This methodology comprehensively integrates historical flight parameters, onboard test data, and fault or maintenance records, combining system functional modeling with advanced data-driven techniques. The approach begins with a detailed analysis of historical data to identify critical systems and predominant fault modes, followed by the development of a functional model and the establishment of a custom parameter spectrum. A Euclidean norm-based technique is then employed to systematically select representative samples of system health, facilitating the creation of a precise health assessment model. This model calculates a health index using the Mahalanobis distance metric. To increase the accuracy of fault detection, a health-fault standardization analysis is implemented, and the early warning threshold is determined by identifying the minimum discrepancy in the health index. The standardized health index is then directly correlated to this threshold, providing an interpretable solution for fault monitoring and warning. The effectiveness of the proposed method is validated through its application to data from civil aircraft air conditioning systems, where it successfully detects faults and issues timely warnings. This study provides a critical technical basis for leveraging historical data-driven fault symptom monitoring, proactive risk early warning, and the enhancement of safety in flight tests.