Scholarlink

<p>The uniqueness of Generalized Normal Distribution Optimization (GNDO) consists of local exploitation and global exploration learning for solving optimization. The main limitation of GNDO tends to get stuck in local optima and has low exploration capability. We propose an Enhanced GNDO (EGNDO) to address these concerns regarding the inversion of geomagnetic anomalies in composite modality. The EGNDO replaces the local exploitation operator of GNDO by integrating Gaussian bare-bones (GB) and differential evolution (DE) operators to effectively balance exploration and exploitation capabilities. The rank-based approach is utilized to select between global exploration and local exploitation operators learning in EGNDO. A chaotic elite learning approach is also used to improve the best solution quality. Before inverting geomagnetic anomalies, sensitivity analyses were conducted on theoretical models with multiple sources, revealing varying degrees of parameter sensitivity. Consequently, an algorithm that balances local exploitation with global exploration is needed to solve this problem. The performance of EGNDO was demonstrated using synthetically generated geomagnetic data (noise-free and noise-added) and three field geomagnetic&#xa0;anomalies with different geological settings collected from some exploration fields in the USA&#xa0;and India for mineral exploration to show the effectiveness in interpreting the data. The EGNDO results are compared with those of standard GNDO, showing that EGNDO performs better&#xa0;in terms of&#xa0;convergence properties, accuracy, robustness, exploration capability, and solution stability in geomagnetic data inversion. Robustness ability analysis is employed to better understand the reliability of the solutions obtained using both algorithms, and the results show that the EGNDO outperformed GNDO in estimating model parameters from geomagnetic anomalies. An exploration capability is required to provide some model parameters for predicting geomagnetic data, which is applied to estimate the uncertainty of model parameters, especially for field geomagnetic anomalies disturbed by noise. Field geomagnetic anomaly applications demonstrated that EGNDO is effective for estimating model parameters from geomagnetic anomalies of ore deposits that are in good concordance with those obtained from published literature and/or drilling information.</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

Enhanced generalized normal distribution optimization for interpreting magnetic data caused by subsurface mineral deposits

  • 期刊论文

  • 发表日期：

    2025年6月20日

查看全文
  • S. Sungkono,
  • Juan P. G. N. Rochman,
  • Saifuddin,
  • Hendra Grandis,
  • Dheo Callisto Furi,
  • Khalid S. Essa,
  • Arkoprovo Biswas

摘要

The uniqueness of Generalized Normal Distribution Optimization (GNDO) consists of local exploitation and global exploration learning for solving optimization. The main limitation of GNDO tends to get stuck in local optima and has low exploration capability. We propose an Enhanced GNDO (EGNDO) to address these concerns regarding the inversion of geomagnetic anomalies in composite modality. The EGNDO replaces the local exploitation operator of GNDO by integrating Gaussian bare-bones (GB) and differential evolution (DE) operators to effectively balance exploration and exploitation capabilities. The rank-based approach is utilized to select between global exploration and local exploitation operators learning in EGNDO. A chaotic elite learning approach is also used to improve the best solution quality. Before inverting geomagnetic anomalies, sensitivity analyses were conducted on theoretical models with multiple sources, revealing varying degrees of parameter sensitivity. Consequently, an algorithm that balances local exploitation with global exploration is needed to solve this problem. The performance of EGNDO was demonstrated using synthetically generated geomagnetic data (noise-free and noise-added) and three field geomagnetic anomalies with different geological settings collected from some exploration fields in the USA and India for mineral exploration to show the effectiveness in interpreting the data. The EGNDO results are compared with those of standard GNDO, showing that EGNDO performs better in terms of convergence properties, accuracy, robustness, exploration capability, and solution stability in geomagnetic data inversion. Robustness ability analysis is employed to better understand the reliability of the solutions obtained using both algorithms, and the results show that the EGNDO outperformed GNDO in estimating model parameters from geomagnetic anomalies. An exploration capability is required to provide some model parameters for predicting geomagnetic data, which is applied to estimate the uncertainty of model parameters, especially for field geomagnetic anomalies disturbed by noise. Field geomagnetic anomaly applications demonstrated that EGNDO is effective for estimating model parameters from geomagnetic anomalies of ore deposits that are in good concordance with those obtained from published literature and/or drilling information.