2023-05-10计算机应用,JournalofComputerApplications2023,43(5):1641-1646ISSN1001-9081CODENJYIIDUhttp://www.joca.cn基于事件触发机制的神经网络同步控制葛超1,2*,常晨蕾1,姚征1,苏皓1,2(1.华北理工大学电气工程学院,河北唐山063210;2.华北理工大学唐山市半导体集成电路重点实验室,河北唐山063210)(∗通信作者电子邮箱gechao365@126.com)摘要:对于混合时滞神经网络同步控制中控制器存在随机摄动的问题,提出一种基于事件触发机制的非脆弱控制器。首先,使用一个服从伯努利分布的随机变量描述控制器增益摄动存在的随机性;其次,在神经网络同步控制过程中引入事件触发机制;接着,构造一个新颖的双边李雅普诺夫函数,充分考虑系统状态信息,同时利用改进的积分不等式对泛函导数进行放缩,得到同步误差系统指数稳定性的充分条件;最后,基于解耦技术设计了非脆弱控制器。通过仿真实例验证了所提控制器的有效性。实验结果表明,在四罐系统中,在相同采样周期下,与现有的指数衰减系数相比,所提控制器获得的指数衰减系数提升了0.16。关键词:神经网络;事件触发机制;指数同步;非脆弱控制器;李雅普诺夫函数中图分类号:TP273文献标志码:ASynchronouscontrolofneuralnetworkbasedonevent-triggeredmechanismGEChao1,2*,CHANGChenlei1,YAOZheng1,SUHao1,2(1.CollegeofElectricalEngineering,NorthChinaUniversityofScienceandTechnology,TangshanHebei063210,China;2.TangshanKeyLaboratoryofSemiconductorIntegratedCircuits,NorthChinaUniversityofScienceandTechnology,TangshanHebei063210,China)Abstract:Concerningtheproblemofrandomperturbationofcontrollerinsynchronouscontrolofneuralnetworkwithmixeddelays,anon-fragilecontrollerbasedonevent-triggeredmechanismwasproposed.Firstly,arandomvariableobeyingBernoullidistributionwasusedtodescribetherandomnessoftheexistenceofcontrollergaindisturbance.Secondly,theevent-triggeredmechanismwasintroducedinthesynchronouscontrolprocessoftheneuralnetwork.Next,anovelbilateralLyapunovfunctionwasconstructedtofullyconsiderthesystemstatusinformation,whilethefunctionalderivativeswerescaledbyanimprovedintegralinequalitytoobtainsufficientconditionsfortheexponentialstabilityofthesynchronizationerrorsystem.Finally,anon-fr...