高速公路匝道回归神经网络控制器的设计

Design of freeway ramp controller based on recurrent neural network

提出用回归神经网络进行入口匝道控制的思路。阐述了Elman回归神经网络原理与入口匝道控制原理,选取上、下游时间占有率和车速作为匝道控制器的输入量,并设计了Elman回归神经网络入口匝道控制器,采用一种改进的算法对回归神经网络进行训练。仿真实验表明,该控制器学习误差小,泛化能力好,具有良好的应用前景。

The idea of recurrent neural network is proposed for on-ramp control. The principles of Elman recurrent neural network and on-ramp control are formulated. Then the occupancy and speed measured in the upstream and downstream portion of a freeway are selected as input variables, and the on-ramp controller based on Elman recurrent neural network is designed. An improved algorithm is used to train the neural network. Simulation experiments show that this controller has such advantages as small learning error and good generalization ability. It is found to be potentially applicable in practice.