【反应扩散Hopfield神经网络的稳定性及应用】赵新泉.pdf

ADissertationSubmittedinPartialFulfillmentoftheRequirementsforthe DegreeofDoctorofPhilosophyinEngineering StabilityandApplicationofHopfieldNeuralNetworks with Reaction-Diffusion Ph.D.Candidate:ZhaoXinquan Major SystemsScience Supervisor ProfessorLiaoXiaoxin Huazhong University of Science Technolo

华中科技大学博士学位论文近稳定性,且在L2模下,利用Halanay微分不等式,得到了的稳态解的局部和全局指数稳定性的充分性判据人们在研究中发现,人工神经网络在实际运行中可产生时滞现象,而变时滞现象更为普遍.因而,变时滞Hopfield神经网络更具有实际意义.本文也重新探讨了这一问题,得到了与以前不同的几个稳定性充分条件.考虑到反映扩散和变时滞三种现象,在L模下,本文探讨了反应扩散变时滞的Hopfield神经网络,得到了几个局部和全局指数稳定性的充分性判据。

华中科技大学博士学位论文 On thebasisof this,the existence and uniqueness of the steady state solutions of Model H-Kneuralnetworkarefurtherdiscussed ApplyingLyapunov function,the author studies the localasymptotic stability and globalasymptotic stabilityof some steadystate solutionsoftheneuralnetworks.