基于神经网络的公交客流预测

公交客流量具有动态性，受多种因素的影响，不能或无法用精确的数学模型进行预测。通过对公交客流量预测的Elman和BP神经网络的建立、学习和训练。并以前三年的公交客流量、国内生产总值、工业总产值、城市人口数作为两种神经网络的输入神经元，第四年的公交客流量作为输出神经元，同时以合肥市公交客流量为例进行分析，结果表明：所建的Elman模型比EBP模型的预测精度高，效果好。

Forecast of Common Traffic Passenger Volume Based on Neural Network

Common traffic passenger volume is dynamic, influenced by multiple factors and is usually unable to be described with accurate mathematical models. BP network and Elman network of passenger volume are built, learned and trained. The common traffic passenger volume, GDP and urban population of the first three years are the input nerve cells of both neural networks, then output the fourth year＇s common traffic passenger volume. A forecast of passenger volume of Hefei is analyzed. The results show that the established Elan model is superior to BP model in forecast accuracy and effect.