基于混沌神经网络的驾驶员动态路径诱导算法研究

Dynamic Route Guidance Algorithm for Driver Based on Transiently Chaotic Neural Network

动态路径诱导旨在向驾驶员提供基于实时交通信息的最佳行驶路径来达到诱导出行的目的,以保证车辆在路网上运行的总费用最小,为驾驶员提供较合理的高效行驶路线.动态路径诱导必须实时保证全局准最优,本文将混沌神经网络应用于动态路径诱导,通过在HNN中引入混沌动态,利用其遍历性进行随机搜索,再由退火策略控制混沌动态逐渐消失并转入HNN进一步优化,从而可保证网络收敛到一个最优或近似最优的稳定平衡点.仿真分析表明:将混沌神经网络应用于动态路径诱导系统中求解最佳路径,总能保证网络收敛到全局最优,同时可有效克服Hopfield神经网络易陷入局部最优解的缺点,具有更高的搜索效率.对于求解连续变量的非线性优化问题提供了一种有效方法,验证了混沌神经网络在动态路径诱导中的有效性.

Dynamic Route Guidance is a system which guides the behavior of drivers by providing optimal route based on real-time traffic information to guarantee the total expenses of traveling in the network is least and provide more reasonable and more efficient driving route for drivers. Dynamic route guidance must ensure global optimum. This paper applies the chaotic neural network to Dynamic Route Guidance, By importing chaotic dynamic charac- teristic in HNN, using its all-over characteristic to search randomly and the annealing algorithm to control and weaken the chaos, and then turning to HNN to be optimized, we can make sure the net convergence to a balanced steady point. Result of simulations shows that the application of the TCNN algorithm to Dynamic Route Guidance System to solve the best path problem can always converge to the globally optimal solution and has higher capability for searching than HNN, and provide an effective way to figure out continuous nonlinear optimizations, thus efficiency of TCNN algorithm used in Dynamic Route Guidance System is proved.