1.华北水利水电大学水资源学院,河南 郑州 450046
2.河海大学水灾害防御全国重点实验室, 江苏 南京 210098
3.河海大学水利水电学院,江苏 南京 210098
谷汶静(2006—),女,主要从事智慧水利、多源水利信息融合及智能预报等工作。
苏怀智(1973—),男,博士,教授,主要从事涉水工程安全防控与提能延寿等工作。
收稿:2025-10-30,
修回:2025-12-19,
录用:2025-12-22,
网络首发:2026-01-27,
纸质出版:2026-07-25
移动端阅览
谷汶静,方正,苏怀智.基于特征筛选和自适应优化的大坝变形预测模型[J].人民珠江,2026,47(7):73-84.
GU Wenjing,FANG Zheng,SU Huaizhi.Dam Displacement Prediction Model Based on Feature Screening and Adaptive Optimization[J].PEARL RIVER,2026,47(07):73-84.
谷汶静,方正,苏怀智.基于特征筛选和自适应优化的大坝变形预测模型[J].人民珠江,2026,47(7):73-84. DOI: 10.3969/j.issn.1001-9235.2026.07.007.
GU Wenjing,FANG Zheng,SU Huaizhi.Dam Displacement Prediction Model Based on Feature Screening and Adaptive Optimization[J].PEARL RIVER,2026,47(07):73-84. DOI: 10.3969/j.issn.1001-9235.2026.07.007.
利用监测数据对大坝变形进行高精度的分析与预测是及时掌握大坝服役状态的重要手段。传统分析预测模型易受到冗余信息影响而难以有效反映大坝变形特征,引入交叉验证的递归特征选择法移除冗余特征,选取最优因子集合从而增强可解释性;利用混沌映射、非线性收敛因子等方法对传统灰狼算法进行多策略改进以期提高全局搜索能力,避免陷入局部最优;采用改进的灰狼算法对梯度提升树的6个参数的最优组合进行全局搜索,避免人为主观因素的影响,以期提高模型预测精度与稳定性。以某工程为例,结果表明该模型所得RMSE为0.287 5,MAE为0.243 3,
R
2
为0.992 1,与各对照模型相比,评价指标均为最优。模型确定了最优因子集合,有效避免了冗余特征的不利影响;借助优化算法的改进与融合,模型有效捕捉了大坝真实变形特征,性能较好,精度较高。
High-precision analysis and prediction of dam deformation utilizing monitoring data is an important means to timely grasp the service status of the dam. The reason is that deformation
as an important indicator comprehensively reflecting whether its structural state is normal or not
is the most intuitive response of the dam structure under the action of loads. Due to the complex operating environment of the dam
numerous factors may affect its deformation state. However
traditional analysis and prediction models have not conducted research on the essential factors influ
encing dam deformation
making them vulnerable to redundant information and thus difficult to effectively reflect the deformation characteristics of dams. Therefore
the recursive feature selection method with cross-validation was taken as the framework and combined with decision tree
ridge regression
and Nu-support vector machine. In the process of the repeated iterative operation
the importance of features was taken as the judgment for removing redundant features and selecting the optimal factor set. This step can further determine the key factors that affect the deformation of the dam
thereby enhancing interpretability. The traditional gray wolf algorithm
inspired by the hunting mechanism of wolves
has the characteristics of strong adaptability and high flexibility. However
it has problems such as a rapid decline in population diversity when facing complex problems. Therefore
multiple strategies including chaotic mapping and nonlinear convergence factors were utilized to conduct improvements on the traditional gray wolf algorithm to enhance the global search ability and avoid falling into a local optimum. The gradient boosting decision tree algorithm based on the tree model for iterative training can specifically mine data features. However
due to its numerous parameters
it is difficult to guarantee the validity of the results based on empirical settings. Therefore
with the aim of improving the prediction accuracy and stability of the model
the improved grey wolf algorithm was adopted to conduct a global search for the optimal combination of the six parameters of the gradient boosting decision tree to avoid the influence of human subjectivity. Taking a certain project as an example
through the screening of the key factors affecting the deformation of the dam
the physical mechanism of the deformation was further clarified. In addition
the dimension of the input set of the prediction model was streamlined
which has improved the accuracy of the model to a certain extent. To quantitatively evaluate
the excellent performance of the model proposed in this paper
an evaluation system was composed of multiple widely adopted indicators
with the aim of comprehensively assessing the model. The results show that the root-mean-square error (RMSE) obtained by the model developed in this paper is 0.287 5; the mean absolute error (MAE) is 0.243 3
and the
R
2
is 0.992 1. Several algorithms that have been widely applied in other fields and achieved good results are selected as control models. The results show that the evaluation indicators of the model proposed in this paper are all optimal
and the prediction results are more consistent with the measured values
which can reflect the deformation trend of the dam. Compared with traditional models
the model proposed in this paper has certain advantages. By determining the optimal factor set
the adverse effects of redundant features are effectively avoided. Through multi-strategy improvement and effective integration of the optimization algorithm
the established prediction model effectively captures the real deformation characteristics of the dam
with good performance and high accuracy.
陈羡濠 , 胡昱 , 王亚军 , 等 . 大坝变形极端梯度提升区间预测模型应用研究 [J]. 水力发电学报 , 2024 , 43 ( 10 ): 121 - 136 .
Chen Xianhao , Hu Yu , Wang Yajun , et al . Dam deformation interval prediction model based on XGBoost [J]. Journal of Hydroelectric Engineering , 2024 , 43 ( 10 ): 121 - 136 . (in Chinese)
周仁练 , 苏怀智 , 韩彰 , 等 . 混凝土坝变形的长期预测模型与应用 [J]. 水力发电学报 , 2021 , 40 ( 9 ): 122 - 131 .
Zhou Renlian , Su Huaizhi , Han Zhang , et al . Long-term deformation prediction model of concrete dams and its application [J]. Journal of Hydroelectric Engineering , 2021 , 40 ( 9 ): 122 - 131 . (in Chinese)
琚海荣 , 梁昱晨 , 晁阳 , 等 . 融合鲸鱼算法与Transformer的乌溪江支墩坝变形预测模型研究 [J]. 人民珠江 , 2026 , 47 ( 4 ): 43 - 51 .
Ju Hairong , Liang Yuchen , Chao Yang , et al . Deformation prediction of Wuxijiang Buttress Dam based on whale optimization algorithm and transformer [J]. Pearl River , 2026 , 47 ( 4 ): 43 - 51 . (in Chinese)
张宏瑞 , 曹昕 , 江超 , 等 . 基于注意力机制与深度学习的混凝土坝变形预测模型 [J]. 人民珠江 , 2026 , 47 ( 2 ): 109 - 117 .
Zhang Hongrui , Cao Xin , Jiang Chao , et al . Concrete dam deformation prediction model based on attention mechanism and deep learning [J]. Pearl River , 2026 , 47 ( 2 ): 109 - 117 . (in Chinese)
林智艳 . 基于BP神经网络的土石坝变形预测模型研究 [J]. 人民珠江 , 2020 , 41 ( 6 ): 74 - 78 .
Lin Zhiyan . Research on prediction model of earth-rockfill dam deformation based on BP neural network [J]. Pearl River , 2020 , 41 ( 6 ): 74 - 78 . (in Chinese)
鄢涛 , 陈波 , 曹恩华 , 等 . 基于EEMD-ELM的大坝变形预测模型 [J]. 长江科学院院报 , 2020 , 37 ( 11 ): 70 - 73,80 .
Yan Tao , Chen Bo , Cao Enhua , et al . Prediction of dam deformation using EEMD-ELM model [J]. Journal of Changjiang River Scientific Research Institute , 2020 , 37 ( 11 ): 70 - 73, 80 . (in Chinese)
吴敏妍 , 苏怀智 , 杨立夫 . 混凝土坝变形预测模型的PCA-WOA-LSSVM组合建模方法 [J]. 水电能源科学 , 2022 , 40 ( 9 ): 111 - 114 .
Wu Minyan , Su Huaizhi , Yang Lifu . PCA-WOA-LSSVM combined modeling method for deformation prediction model of concrete dam [J]. Water Resources and Power , 2022 , 40 ( 9 ): 111 - 114 . (in Chinese)
Ren Qiubing , Li Mingchao , Kong Ting , et al . Multi-sensor real-time monitoring of dam behavior using self-adaptive online sequential learning [J]. Automation in Construction , 2022 , 140 : 104365 .
Wu Zixuan , Li Ling , Xu Tingting , et al . Elucidating the multifaceted roles of GPR146 in non-specific orbital inflammation: a concerted analytical approach through the prisms of bioinformatics and machine learning [J]. Frontiers in Medicine , 2024 ( 11 ): 51 - 59 .
Liu Caide , Liu Chao , Wu Jun , et al . Machine learning and weighted gene co-expression network analysis identify a three-gene signature to diagnose rheumatoid arthritis [J]. Frontiers in Immunology , 2024 ( 15 ): 311 - 320 .
Zhang Chao , Zhai Wenyu , Ma Yuyu , et al . Integrating machine learning algorithms and multiple immunohistochemistry validation to unveil novel diagnostic markers based on costimulatory molecules for predicting immune microenvironment status in triple-negative breast cancer, Frontiers in Immunology [J]. Frontiers in Immunology , 2024 ( 15 ): 42 - 59 .
邓思源 , 周兰庭 , 柳志坤 . 基于卡尔曼滤波、分形和LSTM的大坝变形趋势分析方法 [J]. 水利水电科技进展 , 2022 , 42 ( 5 ): 121 - 126 .
Deng Siyuan , Zhou Lanting , Liu Zhikun . Analysis method of dam deformation trend based on Kalman filter, fractal and LSTM [J]. Advances in Science and Technology of Water Resources , 2022 , 42 ( 5 ): 121 - 126 . (in Chinese)
姚佳池 , 赵二峰 , 刘峰 , 等 . 大坝变形的双向门控循环单元网络预测模型 [J]. 水利水运工程学报 , 2025 ( 4 ): 99 - 107 .
Yao Jiachi , Zhao Erfeng , Liu Feng , et al . Prediction model for dam deformation based on bi-directional gated recurrent unit network [J]. Hydro-Science and Engineering , 2025 ( 4 ): 99 - 107 . (in Chinese)
冯子强 , 李登华 , 丁勇 . 基于Blending-Clustering集成学习的大坝变形预测模型 [J]. 水利水电技术(中英文) , 2024 , 55 ( 4 ): 59 - 70 .
Feng Ziqiang , Li Denghua , Ding Yong . Dam deformation prediction model based on blending-clustering ensemble learning [J]. Water Resources and Hydropower Engineering , 2024 , 55 ( 4 ): 59 - 70 . (in Chinese)
白淑英 , 傅志强 , 谢涛 , 等 . 基于机器学习和哨兵2号遥感影像的棉花种植空间分布信息提取 [J]. 江苏农业科学 , 2024 , 52 ( 20 ): 92 - 104 .
Bai Shuying , Fu Zhiqiang , Xie Tao , et al . Extraction of cotton planting spatial distribution information based on machine learning and Sentinel-2 remote sensing imagery [J]. Jiangsu Agricultural Sciences , 2024 , 52 ( 20 ): 92 - 104 . (in Chinese)
王思晨 , 霍彦峰 , 穆溪 , 等 . 基于极端梯度提升树和深度学习方法估算中国地表NO2浓度 [J]. 环境科学学报 , 2023 , 43 ( 10 ): 298 - 308 .
Wang Sichen , Huo Yanfeng , Mu Xi , et al . Estimating ground-level NO2over China by using extreme gradient boosting and deep learning method [J]. Acta Scientiae Circumstantiae , 2023 , 43 ( 10 ): 298 - 308 . (in Chinese)
聂卉 , 吴晓燕 . 结合梯度提升树算法与可解释机器学习模型SHAP的抑郁症影响因素研究 [J]. 数据分析与知识发现 , 2024 , 8 ( 3 ): 41 - 52 .
Nie Hui , Wu Xiaoyan . Detecting depression factors with gradient boosting tree and explainable machine learning model SHAP [J]. Data Analysis and Knowledge Discovery , 2024 , 8 ( 3 ): 41 - 52 . (in Chinese)
李新华 , 崔东文 . 人工电场算法优化的极限学习机大坝变形预测模型研究 [J]. 人民珠江 , 2022 , 43 ( 6 ): 122 - 127 .
Li Xinhua , Cui Dongwen . Dam deformation prediction model based on artificial electric field algorithm-extreme learning machine model [J]. Pearl River , 2022 , 43 ( 6 ): 122 - 127 . (in Chinese)
Zhu Mingyuan , Chen Bo , Gu Chongshi , et al . Optimized multi-output LSSVR displacement monitoring model for super high arch dams based on dimensionality reduction of measured dam temperature field [J]. Engineering Structures , 2022 , 26 ( 8 ): 46 - 56 .
项颂阳 , 许章华 , 张艺伟 , 等 . 高光谱图像分类的ReliefF-RFE特征选择算法构建与应用 [J]. 光谱学与光谱分析 , 2022 , 42 ( 10 ): 3283 - 3290 .
Xiang Songyang , Xu Zhanghua , Zhang Yiwei , et al . Construction and application of ReliEF-RFE feature selection algorithm for hyperspectral image classification [J]. Spectroscopy and Spectral Analysis , 2022 , 42 ( 10 ): 3283 - 3290 . (in Chinese)
马飞虎 , 雷皓安 , 孙翠羽 , 等 . 基于RFE-OPTUNA-XGBoost模型的高速公路逃费模式识别 [J]. 应用科学学报 , 2024 , 42 ( 5 ): 857 - 870 .
Ma Feihu , Lei Haoan , Sun Cuiyu , et al . Highway toll evasion patterns identification based on RFE-OPTUNA-XGBoost model [J]. Journal of Applied Sciences , 24 , 42 ( 5 ): 857 - 870 . (in Chinese)
林小棋 , 任超 , 李毅 , 等 . 基于Relief F-RFE特征优选的桉树人工林提取 [J]. 测绘科学 , 2023 , 48 ( 10 ): 107 - 115 .
Lin Xiaoqi , Ren Chao , Li Yi , et al . Extraction of eucalyptus plantation forest based on Relief F-RFE feature selection [J]. Science of Surveying and Mapping , 2023 , 48 ( 10 ): 107 - 115 . (in Chinese)
殷孝雎 , 都治良 , 卲国策 , 等 . 基于GWO-SVR的风电机组柔性塔架振动建模与仿真分析 [J]. 太阳能学报 , 2023 , 44 ( 8 ): 404 - 411 .
Yin Xiaoju , Du Zhiliang , Shao Guoce , et al . Vibration modeling and simulation analysis of flexible tower of wind turbines based on GWO-SVR [J]. Acta Energiae Solaris Sinica , 2023 , 44 ( 8 ): 404 - 411 . (in Chinese)
俞华 , 刘宏 , 王璇 , 等 . 基于多频超声和GWO-RF算法的绝缘油击穿电压预测研究 [J]. 绝缘材料 , 2025 , 58 ( 1 ): 130 - 136 .
Yu Hua , Liu Hong , Wang Xuan , et al . Research on prediction of breakdown voltage of insulating oil based on multi-frequency ultrasound and GWO-RF algorithm [J]. Insulating Materials , 2025 , 58 ( 1 ): 130 - 136 . (in Chinese)
陈家鑫 , 赵军龙 , 崔文洁 , 等 . 基于GWO-XGBoost算法的流体识别—以陇东油田CX区长2储层为例 [J]. 地球物理学进展 , 2025 , 40 ( 3 ): 1115 - 1124 .
Chen Jiaxin , Zhao Junlong , Cui Wenjie , et al . Fluid identification based on GWO-XGBoost algorithm: taking Chang 2 reservoir in CX area of Longdong oilfield as an example [J]. Progress in Geophysics , 2020 , 40 ( 3 ): 1115 - 1124 . (in Chinese)
金则澎 , 毛峰 , 程乾 , 等 . 梯度提升回归树在千岛湖水体CDOM反演中的应用 [J]. 遥感信息 , 2022 , 37 ( 1 ): 110 - 118 .
Jin Zepeng , Mao Feng , Cheng Qiang , et al . Application of gradient boosting regression tree in CDOM inversion of Qiandao Lake [J]. Remote Sensing Information , 2022 , 37 ( 1 ): 110 - 118 . (in Chinese)
邢昭 , 孟小军 , 袁晶晶 , 等 . 信息量支持下SVM-GBDT模型的滑坡危险性评价 [J]. 科学技术与工程 , 2025 , 25 ( 7 ): 2712 - 2720 .
Xing Zhao , Meng Xiaojun , Yuan Jingjing , et al . Landslides hazard assessment based on SVM-GBDT and gradient boosting decision tree model supported by information quantity [J]. Science Technology and Engineering , 2020 , 25 ( 7 ): 2712 - 2720 . (in Chinese)
马文伟 . 基于特征选择与BO-GBDT的工作面瓦斯涌出量预测方法 [J]. 工矿自动化 , 2024 , 50 ( 12 ): 136 - 144 .
Ma Wenwei . Prediction method of gas emission in working face based on feature selection and BO-GBDT [J]. Journal of Mine Automation , 2024 , 50 ( 12 ): 136 - 144 . (in Chinese)
向华 , 夏文龙 , 刘波涛 , 等 . 基于梯度提升回归树的气井油管积液高度预测 [J]. 长江大学学报(自然科学版) , 2024 , 21 ( 5 ): 94 - 101 .
Xiang Hua , Xia Wenlong , Liu Botao , et al . Prediction of liquid accumulation height in oil pipelines of gas wells using gradient boosting regression trees [J]. Journal of Yangtze University (Natural Science Edition) , 2024 , 21 ( 5 ): 94 - 101 . (in Chinese)
吴中如 . 水工建筑物安全监控理论及其应用 [M]. 北京 : 高等教育出版社 , 2003 .
Wu Zhongru . Safety monitoring theory & its application of hydraulic structures [M]. Beijing : Higher Education Press , 2003 . (in Chinese)
王瑞婕 , 包腾飞 , 李扬涛 , 等 . 基于多因子融合和Stacking集成学习的大坝变形组合预测模型 [J]. 水利学报 , 2023 , 54 ( 4 ): 497 - 506 .
Wang Ruijie , Bao Tengfei , Li Yangtao , et al . Combined prediction model of dam deformation based on multi-factor fusionand Stacking ensemble learning [J]. Journal of Hydraulic Engineering , 2023 , 54 ( 4 ): 497 - 506 . (in Chinese)
0
浏览量
251
下载量
0
CSCD
关联资源
相关文章
相关作者
相关机构
京公网安备11010802024621