关键词: 关键词: 异构多智能体, 超螺旋滑模, 动态事件触发机制, 固定时间有界稳定

Abstract: Abstract: To address the control challenges of aerial-ground heterogeneous formations under strong disturbances and communication constraints, a dynamic event-triggered method integrating RBF neural networks and fixed-time super-twisting sliding mode is proposed. Firstly, based on the second-order dynamic model of the unmanned aerial vehicle-unmanned ground vehicle (UAV-UGV) heterogeneous system, a fixed-power nonlinear sliding surface is constructed to reduce the computational burden of the sliding mode surface. A fixed-time controller is designed by combining this surface with a super-twisting reaching law, and RBF neural networks are employed to compensate for lumped disturbances online. Secondly, a dynamic event-triggering mechanism based on task errors and learning activity is devised, and a positive lower bound for the minimum inter-event interval is proven to strictly avoid the Zeno phenomenon. Finally, the fixed-time bounded stability of the closed-loop system is demonstrated using Lyapunov theory. Simulations show that, compared to benchmark methods, the disturbance rejection capability is improved by 40%-60%; the sliding mode convergence speed is accelerated by 6%–50% with a 5%-15% reduction in chattering; the event-triggered interval is increased by 40% under steady-state conditions and by 80% under disturbances, significantly reducing communication frequency and achieving a substantial enhancement in overall performance.

Key words: Keywords: heterogeneous multi-agent systems, super-twisting sliding mode, dynamic event-triggered mechanism, fixed-time bounded stability