摘要
针对BP神经网络预测方法的稳定性较差,提出运用增加动量项的小波神经网络方法,进行沉降预测,并选出最优的隐含层节点数目,结合某高铁路基沉降监测数据进行实验分析。实验表明,该方法实际可行,稳定性较好并且预测精度高。
In view of the poor stability of the BP neural net- work prediction, the paper puts forward the method of applying the wavelet neural network with added momentum to the subsid- ence prediction and electing out the optimal number of hidden layer notes. And it further conductes the experimental analysis combining with the subsidence data of some highway subgrade, the results of which show that the method have high stability and prediction precision.
出处
《测绘地理信息》
2013年第6期27-29,共3页
Journal of Geomatics
关键词
BP神经网络
小波神经网络
增加动量项
沉降预测
BP neural network
wavelet neural network
add- ed momentum
subsidence prediction
作者简介
丁鸽，硕士生，主要研究工程测量与工程变形监测等。E—mail：645474779@qq．com
