摘要
针对传统统计模型并不能完全涵盖位移影响分量信息以及真实影响分量信息易受到噪声干扰等问题,提出了一种融合小波阈值理论与多维自回归的混凝土坝位移时序预报模型。该方法主要是将小波阈值理论与时间序列算法结合起来创建混凝土坝位移时序预报模型,模型通过不同小波分解层数、小波基、阈值选取准则、阈值函数集成出一个MATLAB编码平台进行数据平滑处理,能高效挖掘大坝位移数据的影响分量信息,并选择自回归(autoregressive model, AR)时间序列模型作为预报模型。实例应用表明,新的融合模型预测性能较好,能有效监测大坝运行状态,且其分析结果对于其他数字工程的数据预测也具参考价值。
Aiming at the problems that the traditional statistical model can not fully cover the displacement influence component information and the real influence component information is susceptible to noise interference,a time series model for concrete dam displacement prediction based on wavelet threshold theory and multi-dimensional autoregressive was proposed.The core of this method was to integrate a MATLAB coding platform for data smoothing through different wavelet decomposition layers,wavelet bases,threshold selection criteria,and threshold functions,so as to fully and efficiently mine the influence component information of dam displacement data,and selected the auto-regressive time series model as the prediction model.The engineering application showed that the new fusion model had better prediction performance,which ensured the safe operation of the dam comprehensively and efficiently,and its analysis results had a certain reference value for the data prediction of other digital projects.
作者
万祥
魏博文
徐富刚
张升
WAN Xiang;WEI Bowen;XU Fugang;ZHANG Sheng(School of Infrastructure Engineering,Nanchang University,Nanchang 330031,China)
出处
《人民长江》
北大核心
2023年第7期203-209,共7页
Yangtze River
基金
国家自然科学基金项目(51869011,52169025)
江西省杰出青年自然科学基金项目(20192ACB21022)
江西省自然科学基金项目(20192BAB216040)。
关键词
小波阈值
多维自回归模型
混凝土重力坝
位移预报
wavelet threshold theory
multi-dimensional autoregressive model
concrete gravity dam
displacement prediction
