城市交通流小波神经网络预测方法研究

On the Prediction of Urban Traffic Flow Using the Wavelet Neural Network

城市交通流具有复杂性、随机性、模糊性和非线性的特点,城市交通流的准确预测对于解决城市交通问题具有现实意义。针对RBF网络存在的隐层节点中心难以求得的问题,通过构建的小波神经网络对扬州市文昌西路时代广场路口东西方向的交通流进行仿真预测。结果表明,小波神经网络可以准确地、实时地预测交通流,与RBF网络的预测结果对比,小波神经网络拥有相对较高的预测精度。

The Urban traffic flow has the characteristics of complexity,randomness,fuzziness and non linearity. The accurate prediction of traffic flow is of practical significance to solving urban traffic problems. The structure of wavelet neural network and its principle are introduced in this article. In view of the problem with hidden layer nodes existing in RBF network,and by constructing the wavelet neural network,the east-west traffic flow at the intersection of Times Square on Wenchang West Road in Yangzhou city is simulated. The experimental result shows that the wavelet neural network can make accurate and real-time prediction of the traffic flow. Contrasted with RBF network,the wavelet neural network has relatively high predictive precision.