Adaptive Dynamic Programing for Optimal Consensus of Multi-agent Systems Using ANFIS
Yinsheng Yu, Songbo Xie, Xianxia Zhang · 2022 International Conference on Advanced Robotics and Mechatronics (ICARM) · 2022
A new adaptive dynamic programming approach is presented in this paper. It proposes ANFIS-based distributed consensus control of linear multi-agent systems. Firstly, the optimal tracking control problem is formed for a multi-agent system. Secondly, a policy iteration-based ADHDP approach is used to compute the performance index and control policy. Moreover, the convergence analysis of the optimal solutions is given. Thirdly, the control policy is approximated using an ANFIS based actor network, while the iterative performance index function is approximated using an NN based critic network, and this algorithm implements the policy iteration approach online without using system dynamic knowledge. Finally, the suggested optimal tracking control approach is demonstrated accroding to simulation results.