基于光滑控制的神经网络给定时间同步

基于一般的状态反馈和时变的增益函数,设计了一类全新的光滑控制协议,研究了神经网络的给定时间主从同步问题。与传统的有限时间控制或固定时间控制不同,该同步时间可以根据任务的需求而预先设定,从而具有更好的实用性。所提出的控制协议的同步时间不依赖于任何系统初值和控制器参数,这与目前大多数有限时间控制理论的同步时间由系统的初值和控制器参数决定有着本质的区别。同时,文中设计的控制协议没有使用非连续或非光滑的状态反馈,减少了理论分析难度。最后,数值仿真验证了理论分析的有效性。

Prespecified-time Synchronization of the Neural Networks via Smooth Controllers

This paper considers the prespecified-time master-slave synchronization of the neural networks. Based on the regular state feedback and the time-varying gain function, this paper designs a new smooth control protocol to achieve the synchronization of the neural networks in a prespecified time. Different from the traditional finite-time or the fixed-time control, the synchronization time can be assigned in advance according to the task requirements, and thus it has the better practicability. Meanwhile, the synchronization time in the proposed protocol is independent of any initial values or controller parameters, which is essentially different from most existing finite-time control methods. Besides, the proposed protocol does not use the discontinuous or non-smooth state feedbacks, which reduces the difficulty in the theoretical analysis. Finally, some numerical simulations confirm the effectiveness of the theoretical analysis.