DMET-Based Fuzzy Optimized Consensus Control for Nonlinear MASs With Quantized Reference
Yang Liu, Xiangpeng Xie, Reinaldo M. Palhares, Jiayue Sun
Abstract
This paper investigates the dynamic memory eventtriggered (DMET) fuzzy optimized consensus control for nonlinear multi-agent systems (MASs) with quantized reference signal. To alleviate the communication burden, a dual communication channels DMET scheme is proposed, which encompasses eventdriven communication for interactions among followers and communication between controllers and actuators. In comparison to the traditional dynamic event-triggered (DET) scheme, the devised DMET scheme incorporates historical information of the dynamic variable, resulting in longer triggering time intervals. Note that the problem of non-differentiability in backstepping method is generated by the event-triggered communication and quantization. To address this challenge, a smooth signal generator is introduced to reconstruct the step signals into the differentiable new one. Meanwhile, a reinforcement learning (RL) approach is employed to optimize the controllers, which utilizes an identifiercritic-actor architecture with fuzzy logic system (FLS) approximations at each step of backstepping method. The effectiveness of the proposed control method is demonstrated through simulations, confirming its capabilities in achieving optimized consensus control while mitigating communication loads.