一类复值时滞中立型神经网络的稳定性分析  

Stability Analysis of a Class of Complex-valued Neutral Neural Networks with Delays

作　　者：肖瑶 段腾飞 XIAO Yao;DUAN Tengfei(School of Applied Mathematics,Nanjing University of Finance and Economics,Nanjing,Jiangsu,China 210046)

机构地区：[1]南京财经大学应用数学学院,江苏南京210046

出　　处：《昆明学院学报》2022年第6期46-53,共8页Journal of Kunming University

摘　　要：中立型神经网络近年来得到广泛关注,然而对于复值以及含有时滞的模型却还未涉及.通过一定的变换,可将系统转换为微分代数神经网络,从而可利用微分代数不等式方法,而不再需要构造复杂的Lyapunov函数,利用全局指数自同步概念得到了更加简洁有效的全局指数稳定性的充分条件.最后,通过数值实例以及MATLAB仿真验证了定理的有效性和可行性.Neutral-type neural networks have received extensive attention in recent years.However, complex-valued and time-delay models have not been studied yet.Through some transformations, the system is transformed into a differential-algebraic neural network so that the differential-algebraic inequality method is used instead of constructing multifarious Lyapunov functions.A more concise and effective sufficient condition for the global exponential stability of the system is obtained using the concept of global exponential self-synchronization.Finally, the validity and feasibility of the theorem are verified by numerical example and MATLAB simulation.

关 键 词：复值时滞中立型神经网络 微分代数不等式 全局指数自同步 全局指数稳定性 

分 类 号：O175[理学—数学] TP183[理学—基础数学]