In this paper, the distributed optimal consensus problem is investigated for a class of continuous-time nonlinear multi-agent systems with input saturation. Non-quadratic cost functions are introduced to handle input constraints and a novel distributed optimal consensus protocol is derived based on event-triggered adaptive dynamic programming method. An online implement scheme is designed under actor-critic network framework in order to obtain the solutions of Hamilton-Jacobi-Bellman equations online. The computation and communication loads are effectively reduced since the weight estimation vectors and controllers are updated only at event-triggered instants. Detailed analysis is presented based on Lyapunov stability theory which guarantees that the weight estimation errors and local consensus errors are uniformly ultimately bounded. Furthermore, it proves that Zeno behaviour can be effectively avoided. Finally, the simulation examples are presented to validate the proposed strategy.

本文研究了一类具有输入饱和的连续时间非线性多智能体系统的分布式最优一致性问题。引入非二次成本函数来处理输入约束，并基于事件触发自适应动态规划方法推导出一种新颖的分布式最优共识协议。为了在线获得Hamilton-Jacobi-Bellman方程组的解，在actor-critic网络框架下设计了一种在线实现方案。由于权重估计向量和控制器仅在事件触发的时刻更新，因此有效地减少了计算和通信负载。基于李雅普诺夫稳定性理论进行了详细的分析，该理论保证了权重估计误差和局部一致性误差最终一致有界。此外，证明了芝诺行为是可以有效避免的。最后，给出了仿真例子来验证所提出的策略。