任务分配问题的瞬态沸混沌神经网络解法

采用具有瞬态混沌特性的神经网络（TCNN）解任务分配问题，该方法利用神经元的自反馈产生混沌动态，由于混沌动态特性具有很强的搜索全局最优妥的能力，有效地避免了传统Hopfield神经网络（HNN）极易陶入局部极小的缺陷，同时利用时变参数控制混沌行为，使网络在经过一个矩暂的混沌倒分岔后逐渐于一般的Hopfield神经网络，保证网络收敛到一个最优或近似最优的稳定平衡点。仿真结果表明，TCNN解任务分配问题时，总能收敛到全局最优或几乎接近全局最优，同时具有更高的搜索效率，另外，还用此方法求解了属于NP－完全问题的实时分布处理系统的任务分配问题。

SOLVING ASSIGNMENT PROBLEMS BY ARTIFICIAL NEURAL NETWORK WITH TRANSIENT CHAOS

Assignment problems are solved by a neural network model with transient chaos (TCNN).Compared with the conventional Hopfield neural network (HNN),TCNN would not be stuck into local minima by introducing chaos,which is generated by negative self feedback into HNN.With a time variant parameter to control the chaos,the TCNN goes through an inverse bifurcation process and gradually approaches to HNN that converges to a stable equilibrium point.Numerical simulations of assignment problems show that the TCNN has higher ability to search for globally optimal or near optimal solution and higher efficiency of searching than HNN.The TCNN is extended for solving assignment problems in a real time distributed system,which is one of the NP complete problems.