Scholarlink

<p>Post-mixed abrasive water jet (AWJ) cutting is widely used in rock excavation, mining, and tunneling for its high efficiency and low thermal impact. However, the lack of accurate cutting depth models limits its use in complex conditions. This study determined RHT constitutive model parameters for limestone via mechanical tests and used a coupled SPH-FEM algorithm to establish and validate a numerical model. A BP neural network model is further applied to improve cutting depth prediction accuracy. Results show that horizontal, vertical, and shear stresses range 1.6, 1.6, and 2.5 times the jet diameter, respectively. Maximum stress values reach 0.468 GPa, 0.629 GPa, and 0.179 GPa, decreasing toward the jet edge, with maximum reductions of 76.07%, 93.48%, and 64.25%, respectively. Cutting depth increases with jet pressure (227.78% increase), but decreases with higher traverse speeds and standoff distances (reductions of 65.9% and 15.45%, respectively). Traverse speed has the greatest effect, followed by jet pressure and standoff distance. Within the scope of this study, the optimal parameters for limestone cutting are 240&#xa0;MPa jet pressure, 120&#xa0;mm/min traverse speed, and 12&#xa0;mm standoff distance. The BP neural network achieved an MSE of 0.2592 mm<sup>2</sup> and an MAE of 0.4163&#xa0;mm in model validation, with over 90% of cases having MAE &lt; 0.5&#xa0;mm; in external validation using additional prediction sets and supplementary experiments, the maximum relative error was 6.4%, indicating reliable prediction performance. The findings support AWJ use in tunneling, mining, and rock fragmentation, and lay the foundation for intelligent AWJ cutting systems.</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

Intelligent Prediction of Cutting Depth and Parameter Optimization in Abrasive Waterjet Cutting

  • 期刊论文

  • 发表日期：

    2025年10月22日

查看全文
  • Zuliang Zhong,
  • Kaixin Zhu,
  • Jiren Tang,
  • Xinrong Liu,
  • Zezhou Li

摘要

Post-mixed abrasive water jet (AWJ) cutting is widely used in rock excavation, mining, and tunneling for its high efficiency and low thermal impact. However, the lack of accurate cutting depth models limits its use in complex conditions. This study determined RHT constitutive model parameters for limestone via mechanical tests and used a coupled SPH-FEM algorithm to establish and validate a numerical model. A BP neural network model is further applied to improve cutting depth prediction accuracy. Results show that horizontal, vertical, and shear stresses range 1.6, 1.6, and 2.5 times the jet diameter, respectively. Maximum stress values reach 0.468 GPa, 0.629 GPa, and 0.179 GPa, decreasing toward the jet edge, with maximum reductions of 76.07%, 93.48%, and 64.25%, respectively. Cutting depth increases with jet pressure (227.78% increase), but decreases with higher traverse speeds and standoff distances (reductions of 65.9% and 15.45%, respectively). Traverse speed has the greatest effect, followed by jet pressure and standoff distance. Within the scope of this study, the optimal parameters for limestone cutting are 240 MPa jet pressure, 120 mm/min traverse speed, and 12 mm standoff distance. The BP neural network achieved an MSE of 0.2592 mm2 and an MAE of 0.4163 mm in model validation, with over 90% of cases having MAE < 0.5 mm; in external validation using additional prediction sets and supplementary experiments, the maximum relative error was 6.4%, indicating reliable prediction performance. The findings support AWJ use in tunneling, mining, and rock fragmentation, and lay the foundation for intelligent AWJ cutting systems.