Scholarlink

<p>The effective flushing of filters is essential for ensuring the efficiency of micro-irrigation systems and preventing emitter clogging. The water separator mesh filter is a novel filtration device installed at the front end of pumps, and its optimal flushing strategy remains undefined. In this study, indoor physical experiments were conducted by varying the inlet flow rate (<i>Q</i>), sediment concentration (<i>S</i>), water separator length (<i>D</i>), screen aperture (<i>L</i>), and flushing scheme. Evaluation indicators included residual sediment mass (<i>M</i>), head loss (<i>h</i><sub><i>w</i></sub>), and water consumption rate (<i>η</i>). Range analysis, variance analysis, a backpropagation neural network (BPNN), the Non-dominated Sorting Genetic Algorithm II (NSGA-II), and the technique for order preference by similarity to an ideal solution (TOPSIS) were applied for data analysis and optimization. This study makes the following contributions: (1) The flushing scheme and presence of a water separator were found to significantly influence flushing performance, while <i>Q</i> and <i>S</i> were identified as the most critical factors affecting <i>M</i>, <i>h</i><sub><i>w</i></sub>, and <i>η</i>. (2) Predictive models for <i>M</i>, <i>h</i><sub><i>w</i></sub>, and <i>η</i> were developed using the backpropagation neural network algorithm, the random forest algorithm, the XGBoost algorithm, the support vector regression model, the Gaussian process regression model, and the multiple linear regression model. (3) Appropriate application scenarios for different flushing schemes were proposed. (4) The optimal operating conditions were determined as follows: for the TW scheme, <i>Q</i> = 12 m<sup>3</sup>·h<sup>–1</sup>, <i>S</i> = 0.5&#xa0;kg·m<sup>–3</sup>, <i>D</i> = 720&#xa0;mm, <i>L</i> = 0.18&#xa0;mm, and a flushing duration of 20–30&#xa0;s; for the CW scheme: <i>Q</i> = 12 m<sup>3</sup>·h<sup>–1</sup>, <i>S</i> = 0.5&#xa0;kg·m<sup>–3</sup>, <i>D</i> = 800&#xa0;mm, and <i>L</i> = 0.18&#xa0;mm. These findings provide a theoretical foundation for optimizing the flushing performance of pre-pump filters. They also offer technical support for the application of water separator mesh filters in irrigation systems, contributing to the advancement of water-saving agriculture.</p> Graphical abstract <p></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

Study on the optimal flushing scheme of a pre-pump water separator mesh filter for micro-irrigation based on a backpropagation neural network and the Non-dominated Sorting Genetic Algorithm-II

  • 期刊论文

  • 发表日期：

    2026年6月23日

查看全文
  • Lingwei Chen,
  • Yumin Yang,
  • Hongfei Tao,
  • Xingchen Guo,
  • Qiao Li,
  • Youwei Jiang

摘要

The effective flushing of filters is essential for ensuring the efficiency of micro-irrigation systems and preventing emitter clogging. The water separator mesh filter is a novel filtration device installed at the front end of pumps, and its optimal flushing strategy remains undefined. In this study, indoor physical experiments were conducted by varying the inlet flow rate (Q), sediment concentration (S), water separator length (D), screen aperture (L), and flushing scheme. Evaluation indicators included residual sediment mass (M), head loss (hw), and water consumption rate (η). Range analysis, variance analysis, a backpropagation neural network (BPNN), the Non-dominated Sorting Genetic Algorithm II (NSGA-II), and the technique for order preference by similarity to an ideal solution (TOPSIS) were applied for data analysis and optimization. This study makes the following contributions: (1) The flushing scheme and presence of a water separator were found to significantly influence flushing performance, while Q and S were identified as the most critical factors affecting M, hw, and η. (2) Predictive models for M, hw, and η were developed using the backpropagation neural network algorithm, the random forest algorithm, the XGBoost algorithm, the support vector regression model, the Gaussian process regression model, and the multiple linear regression model. (3) Appropriate application scenarios for different flushing schemes were proposed. (4) The optimal operating conditions were determined as follows: for the TW scheme, Q = 12 m3·h–1, S = 0.5 kg·m–3, D = 720 mm, L = 0.18 mm, and a flushing duration of 20–30 s; for the CW scheme: Q = 12 m3·h–1, S = 0.5 kg·m–3, D = 800 mm, and L = 0.18 mm. These findings provide a theoretical foundation for optimizing the flushing performance of pre-pump filters. They also offer technical support for the application of water separator mesh filters in irrigation systems, contributing to the advancement of water-saving agriculture.

Graphical abstract