基于BP神经网络的公路客运量预测方法

Prediction Method of Highway Passenger Transportation Volume Based on BP Neural Network

公路客运量是交通科学管理的基础性数据资料,能够反映出公路运输产出成果,对提高公路交通管理层次及建立畅通、高效的公路交通系统,具有重要意义。为提高公路客运量的预测精度,选择与公路客运量相关的主要社会指标(包括公路客运量、汽车保有量、国民总收入、人均GDP、人口总量、城镇居民人均可支配收入、社会消费品零售总额和城市化率),运用灰色关联分析法进行计算分析,最终确定公路客运量影响因子为汽车保有量、人均GDP、人口总量和城市化率。将所确定的因子作为公路客运量的预测指标,建立基于BP神经网络的公路客运量预测模型,并对模型进行了应用测试。结果表明:BP神经网络模型具有较高的精度,最小相对误差为1.1%,平均相对误差为2.78%。

Highway passenger transportation volume is basic data of traffic scientific management and can reflect the results of highway transportation,which is of great significance to improve the road traffic management level and establish a smooth and efficient highway traffic system.In order to improve the forecasting accuracy of highway passenger transportation,the gray correlation method is used to compute and analyze.The main predictors are car ownership,per capita GDP,total population number and urbanization. The prediction model of highway passenger transportation is establish based on BP neural network,and then verified with tests. The results showthat highway passenger transportation can be predicted accurately by the model based on BP neural network.The minimum relative error is 1.1% and the average relative error is 2.78%.