摘要
通过对基本BP算法的分析,提出了一种基于局部权重及阈值调整的改进BP算法.结合该改进算法,讨论了在Matlab中创建基于BP网络的交通运输需求预测模型并使用该模型进行预测的过程.同时,将基于局部权重及阈值调整的改进BP算法和加动量项的自适应学习率BP算法的模型的预测效果进行了比较,比较结果表明前者的预测效果优于后者.
According to the analysis result of basic BP algorithm.a improved algorithm based on partial weight and threshold adjusting is put forward. Combining with this improved algorithm, the process of establishing and using prediction model of transport demand in the MATLAB is discussed. At the same time, the prediction effect of model based on improved BP algorithm and model based on BP algorithm which updates weight and threshold values according to gradient descent momentum and an adaptive learning rate are compared. The resuh shows that prediction effect of the former is better than the later.
出处
《甘肃联合大学学报(自然科学版)》
2007年第1期49-52,共4页
Journal of Gansu Lianhe University :Natural Sciences
关键词
人工神经网络
BP算法
预测
交通运输
artificial neutral network
BP algorithm
prediction
transport