摘要
利用动量BP算法改进了BP神经网络的收敛性,建立了过渡段路基沉降预测模型.该模型可克服传统BP神经网络收敛速度慢、易陷入局部最优等的缺点.结合津秦客运专线路桥过渡段路基沉降实测数据,将该优化模型与传统BP神经网络预测模型进行了对比.计算表明,利用动量BP算法改进的神经网络具有较高的预测精度,同时考虑了多个影响因素,因而具有广阔的应用前景.
With the momentum back-propagation method, this paper improved the convergence oi BP neural network and developed a prediction model for subgrade settlement of transition section. The model overcame the disadvantages of traditional BP neural network, such as slow convergence speed and easy running into local optimum. Based on test data of the transition section between bridge and subgrade in Tianjin-Qinhuangdao high speed railway, this paper compared the optimization model with the traditional BP neural network model. The results indicate that the neural network improved by momentum BP algorithm has higher predictive accuracy, and it can consider multiple influence factors simultaneously. As a consequence, it has broad application prospect.
出处
《北京交通大学学报》
CAS
CSCD
北大核心
2012年第1期52-55,62,共5页
JOURNAL OF BEIJING JIAOTONG UNIVERSITY
基金
中央高校基本科研业务费专项资金资助(2009JBM079)
2010
2011年大学生创新实验项目资助
关键词
动量BP算法
过渡段
沉降预测
神经网络
momentum back-propagation
transition section
prediction of settlement
neural network
