缺资料地区可解释性混合机器学习模型中长期径流预报

由宇军; 白云岗; 卢震林; 张江辉; 曹彪; 李文忠; 余其鹰

doi:10.11988/ckyyb.20240319

PDF(6677 KB)

raybet体育在线院报 ›› 2025, Vol. 42 ›› Issue (7) : 52-59. DOI: 10.11988/ckyyb.20240319

水资源

缺资料地区可解释性混合机器学习模型中长期径流预报

由宇军 ¹ ,
白云岗 ² ,
卢震林 ² ,
张江辉 ² ,
曹彪 ² ,
李文忠 ³ ,
余其鹰 ³

作者信息 +

Medium- Long-Term Runoff Forecasting Using Interpretable Hybrid Machine Learning Model for Data-Scarce Regions

Author information +

文章历史 +

摘要

在无资料地区气象、水文等观测资料缺乏,影响径流预报的准确性,直接影响水文预报和防汛抗旱工作的开展。分析无资料地区现有的降水、气温和径流数据在中长期预报中的适用性,进而实现径流预报非常重要。分别采用卷积神经网络算法(CNN)、双向门控循环神经网络(BiGRU)、注意力机制(Attention)和优化粒子群算法(IPSO)构建CNN-BiGRU-Attention、IPSO-CNN-BiGRU-Attention组合模型,再与门控循环单元模型(GRU)和ABCD水量平衡模型进行对比分析,并在玉龙喀什河进行综合评估,并结合SHAP可解释性机器学习方法探究最优模型中输入特征对径流影响的贡献程度。结果表明:加入降水和气温的组合模型IPSO-CNN-BiGRU-Attention预测精度整体优于CNN-BiGRU-Attention、GRU模型,与实际值能够较好地吻合;随着预见期的增加,提出的组合模型在验证期内均方根误差(RMSE)、平均绝对误差(MAE)、平均绝对百分误差(MAPE)、纳什效率系数(NSE)分别为2.11、1.32 m³/s、73.76%和0.94,并且在前3个月预报精度最高。该方法在缺资料地区月径流预报中具有较好的效果。

Abstract

[Objectives] This study aims to analyze the applicability of existing precipitation, temperature, and runoff data in data-scarce regions, and to develop and evaluate a deep learning hybrid model driven by multi-source information for improving the accuracy of monthly runoff forecasting. [Methods] Based on historical precipitation, temperature, and runoff sequences from the Yulongkashi River, a Convolutional Neural Network-Bidirectional Gated Recurrent Unit-Attention (CNN-BiGRU-Attention) model was developed. An Improved Particle Swarm Optimization (IPSO) algorithm was used to optimize this model, forming the IPSO-CNN-BiGRU-Attention hybrid model. The performance of this model was compared with that of the Gated Recurrent Unit (GRU) model and the ABCD water balance model. [Results] The IPSO-CNN-BiGRU-Attention model that incorporated precipitation and temperature data overall outperformed the CNN-BiGRU-Attention and GRU models, showing better agreement with the observed values. As the predication period increased, the proposed model achieved a root mean square error (RMSE) of 2.11 m³/s, a mean absolute error (MAE) of 1.32 m³/s, a mean absolute percentage error (MAPE) of 73.76%, and a Nash-Sutcliffe efficiency (NSE) coefficient of 0.94. The highest forecast accuracy was observed in the first three months. [Conclusions] The IPSO-CNN-BiGRU-Attention model effectively integrates precipitation, temperature, and runoff information, significantly enhancing the accuracy of monthly runoff forecasts in data-scarce regions. The model demonstrates robust performance across different forecast horizons, particularly suitable for short-term predictions of 1-3 months. This approach offers a practical and reliable tool for hydrological forecasting and flood control/drought management in data-scarce basins.

导出引用

由宇军, 白云岗, 卢震林, 等. 缺资料地区可解释性混合机器学习模型中长期径流预报[J]. raybet体育在线院报. 2025, 42(7): 52-59 https://doi.org/10.11988/ckyyb.20240319

YOU Yu-jun, BAI Yun-gang, LU Zhen-lin, et al. Medium- Long-Term Runoff Forecasting Using Interpretable Hybrid Machine Learning Model for Data-Scarce Regions[J]. Journal of Changjiang River Scientific Research Institute. 2025, 42(7): 52-59 https://doi.org/10.11988/ckyyb.20240319

中图分类号： TV121 (径流)

参考文献

列表( 原文顺序 | 文献年度倒序 | 文中引用次数倒序 ) 可视化分析

[1]	常启昕, 孙自永, 潘钊, 等. 高寒山区河道径流的形成与水文调节机制研究进展[J]. 地球科学, 2022, 47(11): 4196-4209. (CHANG Qi-xin, SUN Zi-yong, PAN Zhao, et al. Stream Runoff Formation and Hydrological Regulation Mechanism in Mountainous Alpine Regions: a Review[J]. Earth Science, 2022, 47(11): 4196-4209.(in Chinese)) 本文引用 [1]

[2]	卫伟, 余韵, 贾福岩, 等. 微地形改造的生态环境效应研究进展[J]. 生态学报, 2013, 33(20):6462-6469. (WEI Wei, YU Yun, JIA Fu-yan, et al. Research Progress in the Ecological Effects of Micro-landform Modification[J]. Acta Ecologica Sinica, 2013, 33(20): 6462-6469.(in Chinese)) 本文引用 [1]

[3]	陈娟, 徐琦, 曹端祥, 等. 基于多因子多模式集成的中长期径流预测模型[J]. 水科学进展,水科学进展, 2024, 35 (3): 408-419. (CHEN Juan, XU Qi, CAO Duan-xiang, et al. Medium and Long-term Runoff Forecasting Model Based on Multi-factor and Multi-mode Integration[J]. Advances in Water Science, 2024, 35(3): 408-419.(in Chinese)) 本文引用 [1]

[4]

余其鹰, 胡彩虹, 白云岗, 等. 新疆洪水预报预警中融雪径流模型应用进展[J]. 干旱区地理, 2023, 46(12):1951-1962.

https://doi.org/10.12118/j.issn.1000-6060.2023.153

摘要

针对全球变化背景下极端升温、暴雪和暖湿化现象以及中国新疆地区融雪洪水灾害风险增大问题，概述了新疆不同类型洪水灾害特征，重点阐述了近年发生频率增加、致灾性强、灾害风险增大，但在新疆未引起重视的融雪洪水的研究进展，对比分析了不同类型融雪径流模型特点和研究现状。综合目前融雪径流模型已有进展和面临的挑战，提出新疆未来研究需考虑融雪径流模型的物理机制和融雪消融过程，以提高预报预警精度。回顾了融雪洪水在新疆的预报预警技术，指出构建高精度预报预警融雪洪水模型所面临的风吹雪、冻土表层雪和雪面雨等关键问题，并提出提升新疆洪水模拟、预报预警、应对突发洪水的综合能力的关键技术，为提升新疆融水洪水预报预警技术提供思路与建议。

(YU

Qi-ying

, HU

Cai-hong

, BAI

Yun-gang

, et al. Application Progress of Snowmelt Runoff Model in Flood Forecasting and Early Warning in Xinjiang[J]. Arid Land Geography, 2023, 46(12):1951-1962.(in Chinese))

本文引用 [1]

[5]	周霞, 周峰. 基于VMD-NGO-LSTM的融雪洪水汛期非平稳性极值径流预测模型及应用[J]. 人民珠江, 2024: 45(6):127-137. (ZHOU Xia, ZHOU Feng. Non-stationary Extreme Runoff Prediction Model and Its Application in Snowmelt Flood Season Based on VMD-NGO-LSTM[J]. China Industrial Economics, 2024: 45(6):127-137.(in Chinese)) 本文引用 [1]

[6]	效文静, 周建中, 杨建华, 等. 基于机器学习的长江干流中长期径流预报[J]. 水电能源科学, 2022, 40(9): 31-34, 26. (XIAO Wen-jing, ZHOU Jian-zhong, YANG Jian-hua, et al. Medium Long-term Runoff Forecasting of Yangtze River Mainstream Based on Machine Learning[J]. Water Resources and Power, 2022, 40(9): 31-34, 26.(in Chinese)) 本文引用 [1]

[7]	ASSAF A M, HARON H, ABDULL HAMED H N, et al. A Review on Neural Network Based Models for Short Term Solar Irradiance Forecasting[J]. Applied Sciences, 2023, 13(14): 8332. 本文引用 [1]

[8]

SHAIKH

A K

, NAZIR

, KHAN

, et al. Short Term Energy Consumption Forecasting Using Neural Basis Expansion Analysis for Interpretable Time Series[J]. Scientific Reports, 2022, 12: 22562.

https://doi.org/10.1038/s41598-022-26499-y

https://www.ncbi.nlm.nih.gov/pubmed/36581655

本文引用 [1] 摘要

Smart grids and smart homes are getting people's attention in the modern era of smart cities. The advancements of smart technologies and smart grids have created challenges related to energy efficiency and production according to the future demand of clients. Machine learning, specifically neural network-based methods, remained successful in energy consumption prediction, but still, there are gaps due to uncertainty in the data and limitations of the algorithms. Research published in the literature has used small datasets and profiles of primarily single users; therefore, models have difficulties when applied to large datasets with profiles of different customers. Thus, a smart grid environment requires a model that handles consumption data from thousands of customers. The proposed model enhances the newly introduced method of Neural Basis Expansion Analysis for interpretable Time Series (N-BEATS) with a big dataset of energy consumption of 169 customers. Further, to validate the results of the proposed model, a performance comparison has been carried out with the Long Short Term Memory (LSTM), Blocked LSTM, Gated Recurrent Units (GRU), Blocked GRU and Temporal Convolutional Network (TCN). The proposed interpretable model improves the prediction accuracy on the big dataset containing energy consumption profiles of multiple customers. Incorporating covariates into the model improved accuracy by learning past and future energy consumption patterns. Based on a large dataset, the proposed model performed better for daily, weekly, and monthly energy consumption predictions. The forecasting accuracy of the N-BEATS interpretable model for 1-day-ahead energy consumption with "day as covariates" remained better than the 1, 2, 3, and 4-week scenarios.© 2022. The Author(s).

[9]	PRASAD S S, DEO R C, DOWNS N J, et al. Very Short-term Solar Ultraviolet—A Radiation Forecasting System with Cloud Cover Images and a Bayesian Optimized Interpretable Artificial Intelligence Model[J]. Expert Systems with Applications, 2024, 236: 121273. 本文引用 [1]

[10]

徐冬梅, 夏王萍, 王文川. 基于黏菌算法优化VMD-CNN-GRU模型的年径流预测[J]. 南水北调与水利科技(中英文), 2022, 20(3): 429-439.

(XU

Dong-mei

, XIA

Wang-ping

, WANG

Wen-chuan

. Annual Runoff Prediction Based on VMD-CNN-GRU Model Optimized by Slime Mould Algorithm[J]. South-to-North Water Transfers and Water Science & Technology, 2022, 20(3): 429-439.(in Chinese))

本文引用 [1]

[11]

王小君, 窦嘉铭, 刘曌, 等. 可解释人工智能在电力系统中的应用综述与展望[J]. 电力系统自动化, 2024, 48(4): 169-191.

(WANG

Xiao-jun

, DOU

Jia-ming

, LIU

Zhao

, et al. Review and Prospect of Explainable Artificial Intelligence and Its Application in Power Systems[J]. Automation of Electric Power Systems, 2024, 48(4): 169-191.(in Chinese))

本文引用 [3]

[12]	BUTT M A, QAYYUM A, ALI H, et al. Towards Secure Private and Trustworthy Human-centric Embedded Machine Learning: an Emotion-aware Facial Recognition Case Study[J]. Computers & Security, 2023,125:103058. 本文引用 [1]

[13]	田烨, 谭伟丽, 王国庆, 等. LSTM变体模型在径流预测中的性能及其可解释性[J]. 水资源保护, 2023, 39(3):188-194. (TIAN Ye, TAN Wei-li, WANG Guo-qing, et al. Performance of Variant LSTM Models in Runoff Prediction and their Interpretability[J]. Water Resources Protection, 2023, 39(3):188-194.(in Chinese)) 本文引用 [1]

[14]	刘洋. 基于改进深度前馈神经网络(DFNN)算法的网络入侵检测系统研究与实现[D]. 沈阳: 辽宁大学, 2023. (LIU Yang. Research and Implementation of Network Intrusion Detection System Based on Improved Deep Feedforward Neural Network (DFNN) Algorithm[D]. Shenyang: Liaoning University, 2023.(in Chinese)) 本文引用 [1]

[15]	项新建, 许宏辉, 谢建立, 等. 基于VMD-TCN-GRU模型的水质预测研究[J]. 人民黄河, 2024, 46(3):92-97. (XIANG Xin-jian, XU Hong-hui, XIE Jian-li, et al. Water Quality Prediction Based on VMD-TCN-GRU Model[J]. Yellow River, 2024, 46(3): 92-97.(in Chinese)) 本文引用 [1]

[16]	谢秀华, 李陶深. 一种基于改进PSO的K-means优化聚类算法[J]. 计算机技术与发展, 2014, 24(2): 34-38. (XIE Xiu-hua, LI Tao-shen. An Optimized K-means Clustering Algorithm Based on Improved Particle Swarm Optimization[J]. Computer Technology and Development, 2014, 24(2): 34-38.(in Chinese)) 本文引用 [1]

[17]	韩鹏飞. 气候水文模型ABCD的应用和参数特征研究[D]. 北京: 中国地质大学(北京), 2016. (HAN Peng-fei. Application and Parameter Characteristics of Climate Hydrological Model ABCD[D]. Beijing: China University of Geosciences, 2016.(in Chinese)) 本文引用 [1]

[18]	管光华, 刘王嘉仪, 陈晓楠, 等. 输水渠系水动力数字孪生模型糙率估计方法[J]. 水科学进展, 2023, 34(6):901-912. (GUAN Guang-hua, LIU Wang-jia-yi, CHEN Xiao-nan, et al. Roughness Estimation Methods of Hydrodynamic Digital Twin Models for Canal Systems[J]. Advances in Water Science, 2023, 34(6): 901-912.(in Chinese)) 本文引用 [1]

[19]	TIAN Q, GAO H, TIAN Y, et al. Runoff Prediction in the Xijiang River Basin Based on Long Short-term Memory with Variant Models and Its Interpretable Analysis[J]. Water, 2023, 15(18): 3184. 本文引用 [1]