Event-Triggered Federated Reinforcement Broad Learning for Intelligent Fault Diagnosis in Uncrewed Helicopter Swarm Systems
提出事件触发联邦强化宽度学习框架，结合联邦学习、强化学习和宽度学习，实现无人直升机群系统在动态环境下的高效、安全故障诊断，实验验证其诊断精度优于现有方法。
The intelligent fault diagnosis of uncrewed helicopter swarm systems is a critical challenge due to the inherent complexity, dynamic operational environments, and high-risk nature of their missions. To address these challenges, this article proposes a novel event-triggered federated reinforcement broad learning (ET-FRBL) framework, designed to enhance the reliability, security, and efficiency of fault diagnosis in uncrewed helicopter swarm systems. The proposed framework integrates federated learning (FL) to ensure data privacy and collaborative learning, reinforcement learning (RL) to enable adaptive decision-making under dynamic conditions, and broad learning system (BLS) to achieve rapid knowledge expansion and fault pattern recognition. Furthermore, an event-triggered mechanism is introduced to optimize communication efficiency and computational resource utilization by enabling local fault diagnosis at each uncrewed helicopter prior to collaborative diagnosis. To validate the proposed approach, a hardware-in-the-loop (HIL) fault simulation platform for uncrewed helicopter swarm systems is developed, capable of emulating a wide range of fault scenarios, including actuator and sensor failures. Extensive experimental results demonstrate the superior performance of the ET-FRBL framework in terms of diagnostic accuracy compared to state-of-the-art methods.