摘要
大坝是复杂的变形系统,其变形表现为动态非线性,也存在混沌现象。为充分利用大坝变形监测位移时间序列,实现位移单变量情况下的准确预测,提出了一种小波和混沌神经网络预测新方法,首先对大坝位移变形监测数据进行小波分解,其次对分解后的平滑信号进行傅里叶函数拟合预测,再对细节信号进行软阈值去噪和混沌神经网络预测,最后将预测信号进行小波重构。通过工程实例对比分析了小波和混沌神经网络预测新方法、神经网络模型、多元回归模型在大坝位移变形预测中的精度。结果表明,小波和混沌神经网络预测方法的预测精度最高,可以应用于大坝变形预测。
Dam is a complex system,which deformation manifested as dynamic nonlinear and may appear chaotic phenomenon in sometimes.In order to make full use of monitoring data and realize the accurate prediction in case of one variable,this paper put forward a new method of forecasting,which used wavelet to decompose monitoring data,Fourier function to fit smooth signal,soft threshold in wavelet to denoising,chaotic neural network to predict detail signal and wavelet to reconstruct forecast signal.The prediction accuracy of the new method,neural network model and multiple regression model in dam displacement deformation were analyzed by engineering examples.It is found that the prediction method of wavelet and chaos neural network is reliable and has high precision.Therefore,this model can be used in dam deforma⁃tion prediction.
作者
康传利
陈洋
张临炜
时满星
顾峻峰
KANG Chuanli;CHEN Yang;ZHANG Linwei;SHI Manxing;GU Junfeng(Guangxi Key Laboratory of Spatial Information and Geomatics,Guilin 541006,China;College of Geomatics and Geoinformation,Guilin University of Technology,Guilin 541006,China)
出处
《人民黄河》
CAS
北大核心
2020年第3期101-104,116,共5页
Yellow River
基金
国家自然科学基金资助项目(41461089,41541032)
广西空间信息与测绘重点实验室基金项目(桂科能1638025-26,163802515,151400720)
广西高校科学技术研究项目(KY2015YB126)
广西研究生教育创新计划项目(YCSW2017155)
关键词
相空间重构
LYAPUNOV指数
小波分解和重构
小波去噪
混沌神经网络预测
大坝变形
phase space reconstruction
Lyapunov index
wavelet decomposition and reconstruction
wavelet denoising
chaotic neural net⁃work prediction
dam deformation
