This paper deals with the fixed-time consensus tracking problem for a class of non-strict feedback multi-agent systems (MASs) with quantized input, full-state constraints, unmodeled dynamics and dynamic disturbances under a directed graph topology. A distributed adaptive dynamic surface control strategy is proposed on the basis of hysteresis quantizer. This strategy is a consensus tracking method that combines adaptive neural network control and fixed-time command filter backstepping. The non-strict feedback systems with full-state constraints can be converted into unconstrained systems by introducing a logarithmic nonlinear mapping. Radial basis function neural networks are utilized to approximate unknown nonlinear continuous functions. Unmodeled dynamics are handled by introducing a dynamical signal. In addition, the errors generated by the fixed-time filter are eliminated using the error compensation signals. The designed controller in this paper avoids the problems of “singularity” and “complexity explosion” under the traditional backstepping design framework, and the design introduces a hysteresis quantizer that can effectively conserve network resources. Results reveal that the consensus tracking error of MASs can converge to a small neighborhood of the origin in fixed time and all signals and states in the closed-loop systems are bounded based on the Lyapunov stability theorem and graph theory. The effectiveness of the method is verified by two simulation examples.

本文研究了有向图拓扑下具有量化输入、全状态约束、未建模动态和动态扰动的一类非严格反馈多智能体系统（MAS）的固定时间共识跟踪问题。提出了一种基于滞环量化器的分布式自适应动态表面控制策略。该策略是一种结合了自适应神经网络控制和固定时间命令过滤器反步的共识跟踪方法。通过引入对数非线性映射，可以将具有全状态约束的非严格反馈系统转换为无约束系统。径向基函数神经网络用于逼近未知的非线性连续函数。通过引入动态信号来处理未建模的动态。此外，利用误差补偿信号来消除由固定时间滤波器产生的误差。本文设计的控制器避免了传统backstepping设计框架下的“奇异性”和“复杂性爆炸”问题，设计中引入了滞后量化器，可以有效节省网络资源。结果表明，MAS 的一致跟踪误差可以在固定时间内收敛到原点的一个小邻域，并且基于 Lyapunov 稳定性定理和图论，闭环系统中的所有信号和状态都是有界的。通过两个仿真算例验证了该方法的有效性。