Scholarlink

<p>Lithofacies identification is a fundamental component of reservoir evaluation. However, complex lithofacies characteristics, pronounced heterogeneity, and redundant or unevenly distributed logging data present substantial challenges to achieving accurate classification. To address these challenges, this study proposes a sparrow search algorithm (SSA)-optimized least squares support vector machine (LSSVM) model (SSA-LSSVM). The model incorporates multi-dimensional logging data (gamma ray [GR], deep laterolog [LLD], acoustic log [AC], density log [DEN], compensated neutron log [CNL], caliper log [CAL], and shallow laterolog [LLS]) as input features and employs an SSA to adaptively optimize LSSVM hyperparameters, thereby improving generalizability and predictive performance. Experimental results on the test set demonstrate that the SSA-LSSVM model attains a lithofacies prediction accuracy of up to 88%. Moreover, it significantly outperforms random forest, classification and regression tree (CART) decision trees, and conventional support vector machine models across multiple evaluation metrics, including accuracy, precision, recall, and F1-score, underscoring its effectiveness in addressing reservoir heterogeneity and data complexity. These findings confirm that integrating SSA with LSSVM effectively mitigates hyperparameter optimization limitations, substantially enhancing model robustness and predictive accuracy. The proposed framework thus provides an efficient and reliable intelligent solution for lithofacies identification in oil and gas reservoirs, contributing to greater precision and efficiency in reservoir evaluation.</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

Lithofacies Identification and Classification Based on Least Squares Support Vector Machine Optimized by Sparrow Search Algorithm

  • 期刊论文

  • 发表日期：

    2026年1月5日

查看全文
  • Chenyang Mu,
  • Zhongjie Guo,
  • Guangxu Wang,
  • Xiaoxiao Zhang,
  • Cheng Qian,
  • Changsong Lin,
  • Yang Chen,
  • Yougong Wang,
  • Wei Wu

摘要

Lithofacies identification is a fundamental component of reservoir evaluation. However, complex lithofacies characteristics, pronounced heterogeneity, and redundant or unevenly distributed logging data present substantial challenges to achieving accurate classification. To address these challenges, this study proposes a sparrow search algorithm (SSA)-optimized least squares support vector machine (LSSVM) model (SSA-LSSVM). The model incorporates multi-dimensional logging data (gamma ray [GR], deep laterolog [LLD], acoustic log [AC], density log [DEN], compensated neutron log [CNL], caliper log [CAL], and shallow laterolog [LLS]) as input features and employs an SSA to adaptively optimize LSSVM hyperparameters, thereby improving generalizability and predictive performance. Experimental results on the test set demonstrate that the SSA-LSSVM model attains a lithofacies prediction accuracy of up to 88%. Moreover, it significantly outperforms random forest, classification and regression tree (CART) decision trees, and conventional support vector machine models across multiple evaluation metrics, including accuracy, precision, recall, and F1-score, underscoring its effectiveness in addressing reservoir heterogeneity and data complexity. These findings confirm that integrating SSA with LSSVM effectively mitigates hyperparameter optimization limitations, substantially enhancing model robustness and predictive accuracy. The proposed framework thus provides an efficient and reliable intelligent solution for lithofacies identification in oil and gas reservoirs, contributing to greater precision and efficiency in reservoir evaluation.