With the development of the low-altitude economy and industrial ecosystem, safety issues related to electric vertical takeoff and landing (eVTOL) vehicles have gradually drawn attention. However, closed-loop control poses challenges of fault masking and propagation for eVTOL fault diagnosis. Specifically, when multiple faults occur in different rotors, it’s hard to detect and isolated from easy-to-measure flight states. Therefore, to improve the multi-fault diagnosis accuracy of rotors without speed sensors under closed-loop control, this paper proposes a physical-data hybrid fault diagnosis method that integrates an unknown input observer-based model with a random forest. Firstly, appropriate flight state variables are selected to construct an observer for generating residual signals. Then, the random forest is introduced. By integrating the flight status data and sensor data, a comprehensive diagnostic model is established, and the dynamic association between the flight status and faults is captured, achieving the multi-fault diagnosis of eVTOL under closed-loop control. Simulation experiments on single-fault and multi-fault diagnosis validate the effectiveness of the proposed method.

Fault Diagnosis,eVTOL,Random Forest,closed-loop control