一类具有扇形死区的分数阶神经网络系统同步  

A class of fractional order neural network systems with sector dead zones for synchronization

作　　者：蒋文芳 刘恒 JIANG Wenfang;LIU Heng(School of Software Engineering,Pass College of Chongqing Technology and Business University,Chongqing 401520,China;School of Mathematics and Physics,Guangxi Minzu University,Nanning 530006,China)

机构地区：[1]重庆工商大学派斯学院软件工程学院,重庆401520 [2]广西民族大学数学与物理学院,广西南宁530006

出　　处：《南通大学学报(自然科学版)》2025年第1期51-57,共7页Journal of Nantong University(Natural Science Edition)

摘　　要：针对阶数位于1~2之间的一类具有扇形死区的分数阶神经网络系统同步问题进行研究。通过变量替换法,将阶数在(1,2)上的分数阶神经网络系统同步问题转化为阶数在(0,1)上的同步问题。由于分数阶神经网络中含有未知非线性函数项,因此,使用Lipschitz连续条件进行逼近,将未知函数转化为已知函数,但存在一定误差。为了使从神经网络跟踪主神经网络,设计了合适的同步控制器。基于Lipschitz稳定性准则,分析了同步误差系统的收敛性。最后,给出相关仿真实例,并通过该实例中主-从神经网络系统及误差系统的图像,验证了所提方法的有效性。Synchronization is examined for a class of fractional order neural network systems with sector dead zones and orders between 1 and 2.Through variable replacement,the synchronization problem for systems with orders in(1,2)is reformulated as one with orders in(0,1).Unknown nonlinear functions within these networks are approximated using Lipschitz continuity conditions,converting them into known functions with a residual error.A synchronization controller is constructed to align the slave neural network with the master neural network.Convergence of the synchronization error system is assessed using the Lipschitz stability criterion.Simulation examples are provided,with graphical depictions of the master-slave neural network system and error system illustrating the method′s performance.

关 键 词：扇形死区 分数阶神经网络 同步控制 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]