摘要
运用小波理论和神经网络理论不同结合方法建立地表变形预测模型。文中先建立了较为普遍的松散型的小波去噪神经网络模型和紧致型的小波神经网络模型,分析了小波去噪和BP神经网络的隐含层节点数选取过程。基于实测数据分析可得:三种模型的预测效果较单一的BP神经网络预测效果更好;基于小波变换的神经网络预测模型的平均绝对百分比误差为0.15,优于另两种模型的预测精度。
In this paper,we first establish wavelet denoising BP neural network model and BP wavelet neural network model,and this paper set up a multi-component forecast model based on wavelet transform is combined with BP neural network.Based on the analysis of the measured data:prediction accuracy of three kinds of model are higher than the single BP neural network model;MAPE of BP neural network prediction model based on wavelet transform is 0.15,is better than the other two kinds of model’s prediction accuracy.Key words:wavelet denoising;
作者
崔腾飞
许章平
刘成洲
李一凡
CUI Tengfei;XU Zhangping;LIU Chengzhou;LI Yifan(College of Geomatics,Shandong University of Science and Technology,Qingdao Shandong 266590,Chin)
出处
《北京测绘》
2018年第3期273-277,共5页
Beijing Surveying and Mapping
关键词
小波去噪
小波变换
BP神经网络
wavelet denoising
wavelet transform
BP neural network