交通出行预测的神经网络模型

A Neural Network Model for Trip Generation Forecasting

针对交通出行集计预测模型的缺陷,结合神经网络在非线性关系映射方面的优势,本文提出了交通出行预测的BP神经网络模型。作者在对BP神经网络的结构和算法进行分析的基础上,研究了交通出行预测BP神经网络模型的影响因素、模型结构和模型数据,并采用实际调查数据对模型进行了检验和应用。研究结果表明模型预测精度较高,既有很强的理论优势和解释性,又有良好的操作性。最后,文章讨论了下一步的研究方向。

Due to the drawbacks of aggregate trip generation forecasting model, this paper brings forward a new model using back-propagation neural network, which has the advantages of non-linear relation mapping. Based on the analysis of the structure and the algorithm of a neural network, the influencing factors, network structure and discrete data were fully researched. Furthermore, the model was verified and applied using the data from a big city＇ s resident trip survey. The result shows that the forecasting model has high precision, which proves its good maneuverability as well as profound theoretical advantages. At last, the future research direction is discussed.