混沌神经网络优化机制研究

Research on Optimal Mechanism of Chaos Neural Network

分析了混沌神经网络的优化机制,研究了具有模拟退火特性的混沌神经网络模型,给出了混沌神经网络的能量函数,以及计算网络Lyapunov指数的方法,从理论上证明了当网络参数满足一定条件时,网络具有混沌性状。在仿真实验中,应用Hopfield网络和混沌神经网络求解信道分配问题。结果表明,混沌神经网络在求解优化问题时具有更强的搜索全局最优解的能力,和更快的收敛速度。

The optimization mechanism of the chaos neural network （CNN） is analyzed through the research on the mathematic model of CNN with simulated annealing characteristics. The energy function of CNN and the method to calculate the Lyapunov exponent of CNN are given. It is proved that CNN can present the chaotic properties when its parameters satisfy certain conditions. The simulation results indicate that CNN has a stronger ability to search global optimal solution and a higher convergence speed than Hopfield network, when they are applied to resolve the channel assignment problem.