摘要
在铁路货运量预测中,为改善传统预测方法数据拟合度不高、外推性不强等问题,提出基于BP神经网络技术的货运量预测模型,该模型采用贝叶斯正则化方法以提高神经网络推广能力。实验比较发现,该模型具有较强的自适应性,其拟合、预测结果优于灰色预测模型GM(1,1)和修正指数回归模型,证实了该方法的可行性和可靠性。
In order to raise the data fitting deg ree and improve the extrapolation ability,the authors put forward the railway freight volume forecast modeI based 0n BP neuraI network technology This modeI adopts the Bayesian regularization method to strengthen the extrapolation ability of neural network The results of experiments show lhis modeIis comparatively more self-adaptive The results of Ihe approximation and forecasting of this model are better than that of the gray forecast model GM(1,1) and the corrected index regression model. All above prove that this model is reliable and feasible
出处
《铁道货运》
2005年第9期15-17,共3页
Railway Freight Transport
关键词
货运量
预测
神经网络
BP算法
Freight Volume
Forecast
Neural Network
BP Algorithm