Scholarlink

<p>Precise urban water demand forecasting is critical for successful water allocation, particularly under climate variability and drought conditions. The performance of four single-component soft computing models—Support Vector Regression (SVR), Gaussian Process Regression (GPR), Least Square Boost (LSBoost), and Stepwise Linear Regression (SWLR)—and three hybrid models consisting of SWLR and SVR, GPR, and LSBoost is assessed in this research. Average daily rainfall data from nine gauging stations in Fukuoka City, Japan, for the years 1991–1996, including 1992 when Japan was hit by a severe drought, were used as input variables. The data were divided into two parts, calibration (70%), and validation (30%) parts, and the model performance was measured in terms of Nash-Sutcliffe efficiency (NSE), Root Mean Square Error (RMSE), Pearson Correlation Coefficient (PCC), and Mean Absolute Percentage Error (MAPE). The findings indicate that all the single models were not performing satisfactorily (NSE &lt; 0.50 in validation) and the hybrid models were performing much better. SWLR-SVR and SWLR-GPR obtained NSE higher than 0.95; RMSE is less than 2512, and 8863, respectively. Comparison showed that hybrid forms improved prediction accuracy up to 81% in calibration and 88% in validation when using stand-alone techniques. The findings testify the power of the hybrid models to estimate nonlinear rainfall-demand relation, and offer a working and practical framework of urban water demand prediction. Such approaches can be used to guide decision making processes on water allocation, planning and drought management.</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

Water Demand Forecasting Based on Multi-Rainfall Gauging Stations using Stand-Alone Soft Computing Techniques with Improved Novel Hybrid Paradigms

  • 期刊论文

  • 发表日期：

    2026年2月20日

查看全文
  • Tarek Merabtene,
  • Abdullahi G. Usman,
  • Berna Uzun,
  • Dilber Uzun Ozsahin

摘要

Precise urban water demand forecasting is critical for successful water allocation, particularly under climate variability and drought conditions. The performance of four single-component soft computing models—Support Vector Regression (SVR), Gaussian Process Regression (GPR), Least Square Boost (LSBoost), and Stepwise Linear Regression (SWLR)—and three hybrid models consisting of SWLR and SVR, GPR, and LSBoost is assessed in this research. Average daily rainfall data from nine gauging stations in Fukuoka City, Japan, for the years 1991–1996, including 1992 when Japan was hit by a severe drought, were used as input variables. The data were divided into two parts, calibration (70%), and validation (30%) parts, and the model performance was measured in terms of Nash-Sutcliffe efficiency (NSE), Root Mean Square Error (RMSE), Pearson Correlation Coefficient (PCC), and Mean Absolute Percentage Error (MAPE). The findings indicate that all the single models were not performing satisfactorily (NSE < 0.50 in validation) and the hybrid models were performing much better. SWLR-SVR and SWLR-GPR obtained NSE higher than 0.95; RMSE is less than 2512, and 8863, respectively. Comparison showed that hybrid forms improved prediction accuracy up to 81% in calibration and 88% in validation when using stand-alone techniques. The findings testify the power of the hybrid models to estimate nonlinear rainfall-demand relation, and offer a working and practical framework of urban water demand prediction. Such approaches can be used to guide decision making processes on water allocation, planning and drought management.