Scholarlink

<p>Air overpressure (AOp) generated by blasting is a significant environmental and safety concern in surface mining. Accurate prediction and effective control of AOp are important for satisfying regulatory requirements and reducing limit disturbance to surrounding areas. In this study, a two-phase hybrid metaheuristic–artificial intelligence approach was developed for AOp prediction and blasting parameter optimization. In Phase 1, a functional link neural network (FLNN) was optimized using three metaheuristic algorithms—coral reefs optimization (CRO)-FLNN, elephant held optimization (EHO)-FLNN, and ant colony algorithm (ACOR)-FLNN—developed and evaluated against benchmark models, namely standalone FLNN, artificial neural network (ANN), and classification and regression tree (CART). Of the tested models, the CRO-FLNN model provided the best predictive performance, with RMSE = 8.361 and <i>R</i><sup>2</sup> = 0.884 on the testing dataset, representing an improvement of up to 40% over the standalone FLNN. In Phase 2, the selected CRO-FLNN model was incorporated into an inverse optimization framework to calculate blasting parameters that minimize AOp. The results of sensitivity analysis showed that powder factor and maximum charge per delay were the most influential variables. The optimized blasting designs led to predicted AOp levels below 120&#xa0;dB for the tested cases, with reductions of up to 18% compared with baseline conditions. Overall, the results indicated that the proposed approach combines reliable predictive performance with effective optimization capability. The framework can be integrated into digital blast planning systems offers a supporting approach for operational decision support, regulatory compliance, and improved environmental and safety management in surface mining operations.</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

A Metaheuristic-Driven Predictive-Optimization Framework for Blast-Induced Air Overpressure: Coral Reefs Optimization and Cuckoo Search Integration

  • 期刊论文

  • 发表日期：

    2026年6月22日

查看全文
  • Qiancheng Fang,
  • Jian Zhou,
  • Hoang Nguyen,
  • Danial Jahed Armaghani,
  • Erkan Topal

摘要

Air overpressure (AOp) generated by blasting is a significant environmental and safety concern in surface mining. Accurate prediction and effective control of AOp are important for satisfying regulatory requirements and reducing limit disturbance to surrounding areas. In this study, a two-phase hybrid metaheuristic–artificial intelligence approach was developed for AOp prediction and blasting parameter optimization. In Phase 1, a functional link neural network (FLNN) was optimized using three metaheuristic algorithms—coral reefs optimization (CRO)-FLNN, elephant held optimization (EHO)-FLNN, and ant colony algorithm (ACOR)-FLNN—developed and evaluated against benchmark models, namely standalone FLNN, artificial neural network (ANN), and classification and regression tree (CART). Of the tested models, the CRO-FLNN model provided the best predictive performance, with RMSE = 8.361 and R2 = 0.884 on the testing dataset, representing an improvement of up to 40% over the standalone FLNN. In Phase 2, the selected CRO-FLNN model was incorporated into an inverse optimization framework to calculate blasting parameters that minimize AOp. The results of sensitivity analysis showed that powder factor and maximum charge per delay were the most influential variables. The optimized blasting designs led to predicted AOp levels below 120 dB for the tested cases, with reductions of up to 18% compared with baseline conditions. Overall, the results indicated that the proposed approach combines reliable predictive performance with effective optimization capability. The framework can be integrated into digital blast planning systems offers a supporting approach for operational decision support, regulatory compliance, and improved environmental and safety management in surface mining operations.