用具有混沌特性的神经网络解任务分配问题

利用由一对相互藉合的混沌吸引子作为神经元构造的混沌神经网络来解决任务分配问题。通过与传统Hopfield人工神经网络解决任务分配问题相比，混沌神经网络具有更强的全局搜索能力和寻优能力。实时分布处理系统任务分配问题实例仿结果表明，该网络解任务分配问题有效地避免了Hopfield人工神经网络极易陷入局部极小的缺陷，并具有更高的搜索效率。

Solving assignment problems with chaotic neural network

Assignment problems are solved by a chaos neural network (CNN) in whic h a pair of coupled chaos oscillators act as a neuron. Compared with the convent ional Hopfield neural networks(HNN), CNN can be expected to have higher ability of searching for globally optimal or near-optimal solution to assignment proble m s. Numerical simulations of assignment problems in a real-time distributed syst em shows that CNN can overcome HNNs main drawbacks that suffer from the local minimum and perform even more efficient searching when to solve assignment problems.