Abstract: A short-term prediction method based on wavelet analysis and BP neural network was proposed to meet the short-term prediction demand of traffic zone’s generated traffic. The dbN wavelet function was used by the proposed prediction method to decompose the traffic zone’s generated traffic. And the BP neural network was used to predict each waveform with different frequency bands after decomposition in a short time. Finally, the short-term prediction results of the traffic zone’s generated traffic can be obtained by waveform reconstruction. On the basis of constructing the short-term prediction model of comprehensive wavelet analysis and BP neural network, some actually generated traffic data of the traffic zone were collected, and a comparison model of short-term prediction was constructed to test the prediction accuracy of the proposed model. The test results show that the combined prediction model of wavelet analysis and BP neural network has higher accuracy than BP neural network does alone in predicting the short-term generated traffic of the traffic zone.

Key words: traffic engineering; short-term prediction; wavelet analysis; BP neural network; generated traffic

摘要： 针对交通小区生成交通的短时预测需求，提出了综合小波分析和BP神经网络的短时预测方法。预测方法主要利用dbN小波函数对交通小区生成交通进行小波分解，利用BP神经网络对分解后的多频段波形进行短时预测，最后通过波形重构获得交通小区生成交通的短时预测结果。在构建综合小波分析和BP神经网络短时预测模型基础上，采集交通小区的实际交通生成数据，并构建短时预测的对比模型，检验构建模型的预测精度。检验结果表明：在交通小区的生成交通短时预测方面，综合小波分析和BP神经网络的组合预测模型比单独采用BP神经网络进行预测的精度更高。

关键词: 交通工程；短时预测；小波分析；BP神经网络；生成交通