| 798 | 2 | 483 |
| 下载次数 | 被引频次 | 阅读次数 |
为进一步提升水文报汛数据质量管理体系，实现水文报汛数据实时异常监测，拟结合深度学习算法在水文专业的交叉应用，通过分析水文报汛数据的时序特征，结合了卷积神经网络(CNN)和长短期记忆神经网络(LSTM)来增强对特征的处理，同时考虑河道水量传播过程，融合上下游水位、流量数据中隐含的空间关联特征，以实现多步时序预测，挑选长江流域上中下游典型监测站3年的数据构建数据集并进行模拟，结果表明，构建的CNN-LSTM时序预测模型在水位决定系数(R2)达到0.77,流量模拟的决定系数(R2)达到0.84,能较好拟合水文预测结果；在此基础上构建了“预测-残差-判别”一体化的水文数据异常检测框架，在三峡水库的水位模拟中异常值检出率为100%,决定系数(R2)达到0.991,有效实现报汛异常值检测，能为提高流域水文报汛数据质量提供智能算法基础。
Abstract：To further improve the quality management system of hydrological flood reporting data and achieve real-time anomaly monitoring, this study explores the cross-disciplinary application of deep learning algorithms in hydrology. By analyzing the time-series characteristics of hydrological reporting data, a hybrid model combining Convolutional Neural Networks(CNN) and Long Short-Term Memory(LSTM) networks is proposed to enhance feature extraction and processing. Furthermore, considering the physical process of river flow, spatial correlation features embedded in upstream and downstream water level and flow data are integrated to enable multi-step time-series forecasting. A dataset comprising three years of monitoring data from representative stations across the upper, middle, and lower reaches of the Yangtze River was constructed for simulation. The result demonstrate that the proposed CNN-LSTM model achieves a coefficient of determination(R2) of 0.77 for water level prediction and 0.84 for flow simulation, indicating good performance in hydrological forecasting. Building upon this, an integrated anomaly detection framework based on “prediction-residual-discrimination” is developed. In the case study of the Three Gorges Reservoir, the anomaly detection rate reached 100%, with an R2 of 0.991 for water level simulation, effectively enabling the identification of abnormal flood reporting data. This study provides a robust intelligent algorithmic foundation for enhancing the quality of hydrological reporting data in river basins.
[1] 程海云.智慧水文建设思考与长江实践[J].中国水利，2024(18):19-24.
[2] 程海云，熊明，冯宝飞，等.长江现代水文水资源监测预报预警体系[J].水利水电快报，2023,44(7):6-7.
[3] 程海云.推进新时代长江水文高质量发展体系建设构想[J].人民长江，2023,54(1):88-97.
[4] 程海云，张俊.南水北调中线工程水源区水文水资源预报与调度研究[J].中国水利，2024(20):45-51.
[5] 高熠飞，王建平，李林峰.基于柯西分布的水文序列异常值检测方法[J].河海大学学报(自然科学版),2020,48(4):307-313.
[6] 张日新，朱跃龙，万定生，等.基于特征向量的两阶段异常检测方法研究[J].信息技术，2019,43(11):67-71.
[7] 周研来，郭生练，张斐章，等.人工智能在水文预报中的应用研究[J].水资源研究，2019,8(1):1-12.
[8] 倪宪汉，陈浙梁，李欢，等.基于联邦学习的水文遥测数据异常识别与修复[J].浙江工业大学学报，2023,51(6):610-618.
[9] 李泽宇，乔钢柱，张苗苗.基于对偶对抗学习的多维时间序列异常检测[J].中北大学学报(自然科学版),2024,45(2):205-212.
[10] KULANUWAT L,CHANTRAPORNCHAI C,MALEEWONG M,et al.Anomaly detection using a sliding window technique and data imputation with machine learning for hydrological time series[J].Water,2021,13(13):1862.
[11] FARZAD A,GULLIVER T A .Log message anomaly detection with fuzzy C-means and MLP[J].Applied Intelligence,2022,52(15):17708-17717.
[12] 李云霞，姚建国，万定生，等.一种水文时间序列异常模式检测方法研究[J].计算机技术与发展，2019,29(7):159-163.
[13] LIU F,ZHANG L,LI H,et al.A weakly supervised time series anomaly detection method with dual-association discrepancy[J].Signal,Image and Video Processing,2024,18(5):4831-4839.
[14] DING L,FANG W,LUO H,et al.A deep hybrid learning model to detect unsafe behavior:Integrating convolution neural networks and long short-term memory[J].Automation in Construction,2018,86:118-124.
[15] FILHO J A P,NEVES C F,LOPES W T A,et al.Anomaly detection in hydrological network time series via multiresolution analysis[J].Journal of Hydrology,2024,640(000):15.
[16] NGU H C V,LEE K M.CL-TAD:A contrastive-learning-based method for time series anomaly detection[J].Applied Sciences,2023,13(21):11938.
[17] XI H,YUAN Y,DONG H,ZHANG X.Quantifying the effect of land use and land cover changes on spatial-temporal dynamics of water in Hanjiang River Basin[J].Remote Sensing,2024,16(22):4136.
[18] ZHOU J J,LIN C S,HUAN Y C .Decreasing noise in magnetic anomaly detection basing on wavelet denoising[J].Applied Mechanics & Materials,2013,368-370(III):1860-1863.
[19] ZERAATGARI F Z,HAFEZIANZADEH F,ZHANG Y-X,et al.Exploring galactic properties with machine learning:Predicting star formation,stellar mass,and metallicity from photometric data[J].Astronomy & Astrophysics,2024,688:A33.
[20] VLACHAS P R,BYEON W,WAN Z Y,et al.Data-driven forecasting of high-dimensional chaotic systems with long short-term memory networks[J].Proceedings of the Royal Society A Mathematical Physical & Engineering Sciences,2018,474(2213):20170844.
[21] 王婕婷，张泽珑，李飞江，等.基于图神经网络的时序信号异常检测方法[J].西北大学学报(自然科学版),2025,55(2):343-354.
[22] KIM H,KIM H .Contextual anomaly detection for multivariate time series data[J].Quality Engineering,2023,35(4):686-695.
[23] 张怡雯，朱冰.水情信息报送统计考评系统设计与应用[J].水资源研究，2023,12(1):92-97.
[24] 李少恒，严新军，韩克武，等.融合注意力机制的CNN-LSTM模型预测蒸发皿蒸发量[J].人民长江，2025,56(2):75-81.
[25] LIU S,ZHOU B,DING Q,et al.Time series anomaly detection with adversarial reconstruction networks[J].IEEE Transactions on Automatic Control,2023,35(4):14.
[26] 彭玉洁，张冬冬，徐高洪，等.基于VMD-LSTM模型的三峡水库水面蒸发量预测研究[J].人民长江，2024,55(10):110-118.
[27] LEE S Y,JEONG D Y,CHOI J,et al.LSTM model to predict missing data of dissolved oxygen in land‐based aquaculture farm[J].ETRI Journal,2024,46(6):1047-1060.
[28] 霍纬纲，侯振环.基于多尺度卷积自注意力的多维时间序列预测[J].计算机工程与设计，2023,44(4):1250-1258.
[29] MA T,CHEN X.Time series analysis for the adaptive prediction of total phosphorus in the Yangtze River:A machine learning approach[J].Water,2025,17(4):603.
[30] WU G,ZHANG C.Analysis of water quality prediction in the Yangtze River Delta under the River Chief System[J].Sustainability,2024,16(13):5578.
基本信息:
DOI：10.13928/j.cnki.wrahe.2025.S2.026
中图分类号:TP18;P338;TV87
引用信息:
[1]秦洪亮,秦昊,纪国良.基于深度学习的时序预测模型在水文报汛数据异常检测中的应用研究[J].水利水电技术(中英文),2025,56(S2):115-123.DOI:10.13928/j.cnki.wrahe.2025.S2.026.
基金信息:
中国长江三峡集团有限公司科研项目资助(0704218); 国家重点研发计划(2023YFC3209103); 国家自然科学基金项目(U2340211)
2025-09-20
2025-09-20