未知非线性多智能体系统的分布式有限时间自适应神经网络控制  

Distributed Finite-time Adaptive Neural Network Control for Multi-agent Systems with Unknown Nonlinear

作　　者：楚东港 刘烨[1] CHU Dong-gang;LIU Ye(Shanghai University of Engineering Science,Shanghai 201620,China)

机构地区：[1]上海工程技术大学,上海201620

出　　处：《仪表技术与传感器》2023年第5期112-117,126,共7页Instrument Technique and Sensor

基　　金：国家自然科学基金(61703269)。

摘　　要：针对一类具有未知磁滞输入的非线性多智能体系统,提出了一种基于改进命令滤波器的分布式预设性能自适应神经网络渐近一致控制方案。首先,为保证系统的跟踪性能,提出了基于新型性能函数的有限时间Funnel控制方案,使得跟随者与领导者在有限的时间内输出一致。其次,利用径向基神经网络和放缩方法消除了未知非线性的影响。然后,使用命令滤波器解决了传统递归方法的“微分爆炸”难题。理论分析表明:该控制方案消除了未知磁滞输入的影响并且实现了一致性误差渐近收敛至0。最后通过实例仿真验证了文中控制方案的有效性。For a class of nonlinear multi-agent systems with unknown hysteresis input,this paper proposed a distributed preset performance adaptive neural network asymptotic consensus control scheme based on an improved command filter.First,in order to ensure the tracking performance of the system,a finite-time Funnel control scheme based on a novel performance function was proposed,so that the output of the followers and the leader was consistent within a limited time.Second,the effects of unknown nonlinearities were addressed by using radial basis function neural network and scaling methods.Then,the“differential explosion”difficulty of traditional recursive methods was overcome by using command filters.Theoretical analysis shows that the control scheme not only eliminates the influence of the unknown hysteresis input,but also makes the consensus error converge to zero asymptotically.Finally,the effectiveness of the control scheme in this paper was verified by practical simulation.

关 键 词：自适应控制 径向基神经网络 有限时间Funnel控制 命令滤波器 磁滞输入 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]