Scholarlink

<p>Accurate estimation of reservoir inflow can significantly contribute to proper water resource management and reservoir allocation for downstream demands and hydropower generation. In this research, support vector regression (SVR) was employed to predict the inflow to the Mahabad Dam reservoir in northwestern Iran. When combined with metaheuristic optimization algorithms, machine learning models can lead to highly accurate predictions. Therefore, the novel harris hawks optimization (HHO) algorithm was subsequently utilized to enhance the SVR framework, and the results of the standalone and hybrid models were compared. For this research, monthly precipitation, temperature, and reservoir inflow data with one to three delays over a 28-year (1995–2022) statistical period were utilized as input parameters across six patterns. Modeling with patterns that incorporate a greater number of input parameters yields superior results. Therefore, the sixth pattern, which included all the input parameters, achieved the highest modeling accuracy in both employed models. For this pattern, the hybrid SVR-HHO model demonstrated superior performance, where the root mean square error (RMSE), mean absolute error (MAE), and nash–sutcliffe efficiency coefficient (NSE) values for the test data were 4.39&#xa0;million cubic meters (MCM), 2.7 MCM, and 0.86, respectively. A comparison of the results from the sixth pattern in both the SVR and SVR-HHO models revealed that the HHO evolutionary algorithm successfully enhanced the prediction accuracy. Specifically, for the test data in the sixth pattern, the RMSE and MAE values were reduced by 1.51 MCM and 0.2 MCM, respectively, and the NSE value was increased by approximately 0.13.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhanced Prediction of Reservoir Inflow Via an SVR–Harris Hawks Optimization Hybrid Model: A Case Study of the Mahabad Dam

  • 期刊论文

  • 发表日期：

    2026年4月28日

查看全文
  • Seyed Mohammad Enayati,
  • Shahab Abravesh

摘要

Accurate estimation of reservoir inflow can significantly contribute to proper water resource management and reservoir allocation for downstream demands and hydropower generation. In this research, support vector regression (SVR) was employed to predict the inflow to the Mahabad Dam reservoir in northwestern Iran. When combined with metaheuristic optimization algorithms, machine learning models can lead to highly accurate predictions. Therefore, the novel harris hawks optimization (HHO) algorithm was subsequently utilized to enhance the SVR framework, and the results of the standalone and hybrid models were compared. For this research, monthly precipitation, temperature, and reservoir inflow data with one to three delays over a 28-year (1995–2022) statistical period were utilized as input parameters across six patterns. Modeling with patterns that incorporate a greater number of input parameters yields superior results. Therefore, the sixth pattern, which included all the input parameters, achieved the highest modeling accuracy in both employed models. For this pattern, the hybrid SVR-HHO model demonstrated superior performance, where the root mean square error (RMSE), mean absolute error (MAE), and nash–sutcliffe efficiency coefficient (NSE) values for the test data were 4.39 million cubic meters (MCM), 2.7 MCM, and 0.86, respectively. A comparison of the results from the sixth pattern in both the SVR and SVR-HHO models revealed that the HHO evolutionary algorithm successfully enhanced the prediction accuracy. Specifically, for the test data in the sixth pattern, the RMSE and MAE values were reduced by 1.51 MCM and 0.2 MCM, respectively, and the NSE value was increased by approximately 0.13.