Fault Diagnosis of Unmanned Aerial Vehicle Based on the Slime Mold Algorithm under Imbalanced Data Conditions

With the rapid progress of technology, the application of unmanned aerial vehicles (UAVs) has become widespread in various civilian and military activities. The research on fault diagnosis during a flight is a critical step towards enhancing the safety of UAV systems. Swarm-inspired heuristic algorithms have attracted significant attention owing to their fast iteration speed and strong optimization capabilities. This paper proposes a UAV fault diagnosis method based on the slime mold algorithm (SMA) under the condition of imbalanced data, thus addressing the issue of low diagnostic accuracy caused by a significantly longer normal flight time compared to the fault flight time in UAV operations. Using a real-world dataset of UAV flight faults with multiple fault patterns, the proposed approach optimizes the parameters of the support vector machine through SMA, thereby improving the fault diagnosis performance during actual UAV flights.