本文已被：浏览 487次 下载 232次
投稿时间：2022-06-10 修订日期：2021-11-12
投稿时间：2022-06-10 修订日期：2021-11-12
中文摘要: 针对混凝土拱坝变形机理的复杂性与测值的高度非线性，提出了一种融合残差有效成分的混凝土拱坝变形预测组合模型，解释大坝变形性能。鉴于统计模型无法有效联系筑坝材料性能演变对大坝变形的影响，结合有限元方法计算水压分量，构建混合模型；考虑到混合模型残差序列的混沌与周期性特征，采用极限学习机（ELM）模型和季节性差分自回归积分滑动平均模型（SARIMA），分别对高频与低频信号进行逐一建模预测；考虑到ELM模型参数对模型预测性能的影响，结合具有良好全局搜索能力的粒子群算法（PSO），对其参数寻优，构建了适应于高频信号处理的优化ELM模型；将高低频信号的预测结果与混合模型建模结果叠加，构建混凝土拱坝变形组合预测模型。以某混凝土拱坝为例，通过建立该大坝的3维有限元数值模型，计算位移拱冠梁上的典型测点PLA1、PLA2水压分量，在构建混合模型的基础上，运用组合模型对典型测点的残差序列变化规律进行剖析与预测。分析结果表明：相比于统计模型、混合模型及基于EEMD–PSO–ELM模型，本文所建组合模型的拟合与预测能力更优，有效验证了所建模型的合理性与可行性；同时，该组合模型具有出色的非线性信息挖掘与建模预测能力，可为大坝变形监测数据分析与预测提供技术支撑。
Abstract:For the complexity of deformation mechanism of the concrete arch dam and the highly non-linearity of measured value, the explanatory performance of prediction model for dam deformation is important. A combined forecasting model for concrete arch dam deformation was proposed, which combined the residual effective components. In view of the fact that the statistical model can not effectively relate the influence of material property evolution on dam deformation, the hybrid model was constructed by calculating hydraulic components with the finite element method. At the same time, considering the chaos and periodicity of the mixed model residual sequence, the high-frequency and low-frequency signals were modeled and predicted one by one by using the limit learning machine (ELM) model and the seasonal difference autoregressive integration sliding average model (SARIMA). Considering the influence of ELM model parameters on model predictive performance, an optimized ELM model for high-frequency signal processing was constructed by combining particle swarm optimization (PSO) with good global search ability to optimize its parameters. Finally, the combined forecasting model of concrete arch dam deformation was built by superimposing the forecasting results of high and low frequency signals with the modeling results of the hybrid model. Taking a concrete arch dam as an example, by establishing the three-dimensional finite element numerical model of the dam, the water pressure components of typical measuring points PLA1 and PLA2 on the displacement arch crown beam were calculated. Based on the construction of the mixed model, the residual sequence variation regularity of typical measuring points was analyzed and predicted by the combination model. The analysis results showed that the forecasting results of the combined model were better than those of the statistical model, hybrid model and EEMD–PSO–ELM model, which validated the rationality and feasibility of the model. At the same time, the combined model had an excellent capability of non-linear information mining, modeling and prediction, and could provide new technical support for the analysis and prediction of dam deformation monitoring data.
文章编号：202100544 中图分类号： 文献标志码：
基金项目:国家自然科学基金项目（51869011；52169025）；江西省青年科学重点项目（20192ACB21022）；江西省研究生创新专项资金项目（YC2020-S122）；中国博士后基金项目 (2019M652281)；江西省自然科学基金项目（20192BAB216040）
作者简介:第一作者：魏博文（1981—），男，教授，博士. 研究方向：水工结构与大坝安全监控. E-mail：ncuweibowen@126.com;通信作者：徐富刚, E-mail：xufugang785315056@126.com
引用文本：
魏博文,罗绍杨,徐富刚,袁冬阳,张婉彤.基于监测时序分解再重构的混凝土拱坝位移预测组合模型[J].工程科学与技术,2022,54(5):51-63.
WEI Bowen,LUO Shaoyang,XU Fugang,YUAN Dongyang,ZHANG Wantong.Combined Model of Displacement Prediction for Concrete Arch Dam Based onDecomposition and Reconstruction of Monitoring Time Series[J].Advanced Engineering Sciences,2022,54(5):51-63.
引用文本：
魏博文,罗绍杨,徐富刚,袁冬阳,张婉彤.基于监测时序分解再重构的混凝土拱坝位移预测组合模型[J].工程科学与技术,2022,54(5):51-63.
WEI Bowen,LUO Shaoyang,XU Fugang,YUAN Dongyang,ZHANG Wantong.Combined Model of Displacement Prediction for Concrete Arch Dam Based onDecomposition and Reconstruction of Monitoring Time Series[J].Advanced Engineering Sciences,2022,54(5):51-63.