Abstract In a channel irrigation system, diverse side orifice shapes (e.g., circular, rectangular, or triangular) are widely designed for flow control and regulation. Hence, accurately estimating the discharge diverted from the main channel to the side one is essential for water management. The main objective of this research is to estimate the discharge coefficient ( C d ) for a sharp-crested triangular side orifice under a free flow condition. Three linear data-driven models including locally weighted learning regression (LWLR), multiple linear regressions with interaction (MLRI), and multivariate linear regression (MLR) were developed for this purpose. 570 experimental datasets were used to build the predictive models. Two modeling scenarios (with and without incorporating the upstream flow Froude number) were investigated for estimating C d . The best input combinations for both scenarios were identified by applying the Gamma test. The performance of the models was assessed by using various graphical analysis and statistical metrics. The modeling results indicated that LWLR and MLRI had similar good performance for both modeling scenarios and provided accurate estimation of the C d values. Overall, this study demonstrated the capacities of the data-driven models in estimating the discharge coefficient of triangular side orifice under a free flow condition.

中文翻译：

---

使用数据驱动模型估算自由流动条件下的三角边孔板流量系数

摘要 在渠道灌溉系统中，各种侧孔形状（例如圆形、矩形或三角形）被广泛设计用于流量控制和调节。因此，准确估算从主通道分流到侧通道的流量对于水管理至关重要。本研究的主要目的是估计在自由流动条件下尖顶三角形侧孔的流量系数 (C d )。为此开发了三种线性数据驱动模型，包括局部加权学习回归 (LWLR)、具有交互作用的多元线性回归 (MLRI) 和多元线性回归 (MLR)。570 个实验数据集用于构建预测模型。研究了两种建模方案（包括和不包括上游流量 Froude 数）来估计 C d 。通过应用 Gamma 测试确定了这两种情况的最佳输入组合。通过使用各种图形分析和统计指标来评估模型的性能。建模结果表明，LWLR 和 MLRI 在两种建模场景中都具有相似的良好性能，并提供了对 C d 值的准确估计。总体而言，这项研究证明了数据驱动模型在估计自由流动条件下三角形侧孔板的流量系数方面的能力。建模结果表明，LWLR 和 MLRI 在两种建模场景中都具有相似的良好性能，并提供了对 C d 值的准确估计。总体而言，这项研究证明了数据驱动模型在估计自由流动条件下三角形侧孔板的流量系数方面的能力。建模结果表明，LWLR 和 MLRI 在两种建模场景中都具有相似的良好性能，并提供了对 C d 值的准确估计。总体而言，这项研究证明了数据驱动模型在估计自由流动条件下三角形侧孔板的流量系数方面的能力。