基于小波分析与神经网络的交通流短时预测方法

The Forecasting Approach for Short-term Traffic Flow Based on Wavelet Analysis and Neural Network

摘要: 提出基于小波分析与神经网络的交通流短时预测方法，把多维输入进行小波分解降维，预测由多个子网络独立完成，有效解决了多维神经网络的映射学习容易产生“维数灾”的问题．示例结果表明，该方法比典型的神经网络预测准确度高、误差小．

Abstract: The method based on wavelet analysis and neural network for short-term traffic flow forecasting is presented.The multidimensional inputs are decompounded by wavelet analysis and the forecasting is implemented by several sub neural networks independently.It resolves the dimension-disaster problem effectively in multidimensional neural network mapping.The demonstration results show that the method can evidently decrease prediction error and improve forecasting veracity compared with typical neural network.