基于改进的RBF神经网络的高速公路交通事件检测

根据高速公路有交通事件发生时交通流将产生突变这一原理,采用改进的径向基函数(RBF)神经网络研究高速公路事件检测问题。分析交通流参数在有交通事件发生时的变化规影影响神经网络泛化能力的同时,加入多余节点的删除和合并策略,从而得到精简的网络结构。采用自适应学习方法进行隐含层节点的调整,使网络在不同的训练阶段能够自动选取不同的学习速率。仿真试验表明,该改进算法在高速公路交通事件检测中具有检测率高、学习速度快等优点,具有良好的应用前景。

Freeway Traffic Incident Detection Based on Improved Radial Basis Function Network

According to the principle that an incident on a freeway produces a mutation on traffic flow,freeway incident detection was researched by using improved Radial Basis Function(RBF)Network.The change rule of traffic flow parameters while happening of incident was analyzed.The dynamic neural network architecture was constructed by selecting the proper traffic flow parameters as inputs for pixel level fusion.A strategy of merging and deleting the redundant hidden nodes was introduced to obtain the simplified network structure without prejudice generalization ability of the neural network.The adjustment of hidden nodes was implemented by applying the adaptive learning method to make the neural network automatically select different learning rates in different training stages.Simulation experiments show that this incident detection algorithm has such advantages as high detection rate and fast learning ability.It is found to be potentially applicable in practice.