基于神经动态规划算法的最优路径选择

针对传统动态规划算法在计算大规模路网的优化问题时所表现出来的计算时间长、存储空间大等缺点，引入了一种神经动态规划算法：它将传统的动态规划和BP神经网络结合起来，通过逼近Q学习算法来寻求一种最优策略，最终达到路径优化的目的。将此算法应用于一个交通路网，且用Matlab软件进行仿真，试验表明：该方法的实时性、并行性和全局性都优于传统动态规划，在城市交通流系统中能切实起到路径诱导的作用。

Optimal Route Choice Based on Neuro-Dynamic Programming

Neuro-dynamic programming is introduced to deal with traditional dynamic programming with large storage space and long computing time for optimal route in a gigantic traffic network.By synthesizing the dynamic programming and BP neural network to seek an optimal strategy through the approximate Q-learning algorithm,the purpose of optimizing the path is ultimately achieved.Matlab simulation experiments show that this method of real-time,parallel and the overall situation is superior to the traditional path p...