Scholarlink

<p>The accurate estimation of blast-induced peak particle velocity (PPV) in mining projects is essential for assessing potential structural damage and environmental impact. This study introduces an accurate computational model to estimate the PPV in mining projects. To achieve that, the artificial bee colony (ABC), genetic (GA), grey wolf optimization (GWO), harris hawks optimization (HHO), PSO, and salp swarm optimization (SSO) algorithms were utilized to develop six hybrid least square support vector machine (LSSVM) models. The models were trained and tested on 127 and 23 blasting events, respectively. The capabilities of each model were analyzed using four statistical metrics, including Taylor plots and actual vs. predicted plots. The analysis revealed that the HHO_LSSVM model outperformed the other hybrid LSSVM models with a root mean square error (RMSE) of 0.056 and a performance (R) of 0.9973 in the testing phase. The visual interpretation also confirmed the robustness of the HHO_LSSVM model. In addition, it was observed that the PSO_LSSVM and SSO_LSSVM models overfitted due to weak multicollinearity among the features, which significantly affected PPV estimation. SHAP (SHapley Additive exPlanations) analysis of the blast fragmentation model reveals that maximum charge per delay (Q) has the highest predictive importance at 70%, demonstrating a strong positive linear relationship: higher instantaneous energy input critically enhances fragmentation performance. Finally, this study introduced the HHO_LSSVM model as an optimal performance model to assess accurate PPV in mining projects.</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

Assessment of Blast-Induced Peak Particle Velocity Using Six Hybrid Least Squares Support Vector Machine Models: A Case Study of an Opencast Mine in the Ramagundam Area, Telangana

  • 期刊论文

  • 发表日期：

    2025年11月1日

查看全文
  • Jitendra Khatti

摘要

The accurate estimation of blast-induced peak particle velocity (PPV) in mining projects is essential for assessing potential structural damage and environmental impact. This study introduces an accurate computational model to estimate the PPV in mining projects. To achieve that, the artificial bee colony (ABC), genetic (GA), grey wolf optimization (GWO), harris hawks optimization (HHO), PSO, and salp swarm optimization (SSO) algorithms were utilized to develop six hybrid least square support vector machine (LSSVM) models. The models were trained and tested on 127 and 23 blasting events, respectively. The capabilities of each model were analyzed using four statistical metrics, including Taylor plots and actual vs. predicted plots. The analysis revealed that the HHO_LSSVM model outperformed the other hybrid LSSVM models with a root mean square error (RMSE) of 0.056 and a performance (R) of 0.9973 in the testing phase. The visual interpretation also confirmed the robustness of the HHO_LSSVM model. In addition, it was observed that the PSO_LSSVM and SSO_LSSVM models overfitted due to weak multicollinearity among the features, which significantly affected PPV estimation. SHAP (SHapley Additive exPlanations) analysis of the blast fragmentation model reveals that maximum charge per delay (Q) has the highest predictive importance at 70%, demonstrating a strong positive linear relationship: higher instantaneous energy input critically enhances fragmentation performance. Finally, this study introduced the HHO_LSSVM model as an optimal performance model to assess accurate PPV in mining projects.