时滞型脉冲神经网络的全局指数稳定性.pdfVIP

第 26卷第 1期 大 学 数 学 Vo1．26，№ ．1 201O年 2月 COLLEGE MATHEM ATICS Feb．2010 GlobalStabilityofNeuralNetworkswith DelaysandImpulses YANGFeng-jian， ZHANGChao—long， WuDong—qing， rANGJian-fu， CHEN Xin一1)ling (ComputationalScienceDepartmentofZhongkaiUniversityofAgricultureandEngineering，Guangzhou510225，China) Abstract：Thispaperisconcernedwithaclassofneuralnetworkswithdelaysandinpulses．Sufficient c0nditionsareobtained fortheexistenceand globalexponentialstabilityofa uniqueequilibrium without assumingtheaetivationfunctiontObebounded，monotonicordifferentiable．An illustrativeexampleisgiven todemonstratetheeffectivenessoftheobtainedresults． Keywords：impulses；delays；neuralnetworks；globalexponentialstability CLC Number：TP183 DocumentCode：A Article1D：l672—1454(2010)01-0028—05 l lntroduction ThedynamicsofHopfieldneuralnetworksaroused extensiveresearchesbynumerousscholars duo to its applicability in solving image processing， signalprocessing and pattern recognition problems．In1988，Chuaputforwardthemodelofcellularneuralnetworkswithtimedelayin[1]： z(￡)一～ ()+∑ng，(，(￡一r))+J，，i一1，2，…，， andpres；entedtheelectriccircuitstoutilizethemode1．Ref．[2]introducestheapplicationofthis networkinimageinformationandprocessing．Someauthordiscusseddynamicpropertyofthiskindof neuralnetworksin[3—72．Caoetc．discussedasymptoticstabilityofthemoregeneralneuralnetwork ofDCNNtypein[8一103： 生 一一 ()+∑ng，(，(￡))+∑ g，((￡一))+J，i一1，2，…，m J一 1 ·一 1 Mostwidely used neuralnetworks is neither purely continuous——time nor purely discrete——time ones，these are called impulsiveneuralnetworks． In thispaper，we considerthedynamics ofthe