Scholarlink

<p>The performance of Earth Pressure Balance (EPB) machines strongly depends on the proper conditioning of excavated materials. This research presents a novel gene expression programming (GEP) model to predict the slump value of conditioned marly limestone in the pressure chamber of EPB-TBMs. A total of 97 slump tests were conducted to examine the effects of bentonite slurry on improving the workability of crushed marly limestone. Fine particle percentage (F), bentonite slurry concentration (C<sub>B</sub>), and slurry injection ratio (SIR) were selected as input variables, while slump value (S) served as the output. The GEP model was developed using 78 training and 19 testing data sets, and its performance was compared to a linear multivariable regression (LMR) model. Evaluation of the models were conducted using four statistical metrics: coefficient of determination (R2), variance accounted for (VAF), root mean square error (RMSE), and mean absolute error (MAE). The results showed that GEP outperformed LMR, achieving R2 = 0.913, VAF = 89.2%, RMSE = 2.294, and MAE = 1.98 in the test phase. Moreover, residual analysis was performed and showed that residuals were normally distributed and randomly scattered around zero, confirming no systematic bias or overfitting. Finally, a sensitivity analysis revealed that SIR had the greatest impact on slump prediction, while F had the least. The GEP model developed in this study could be utilized as a reliable tool to more accurate prediction of the slump value to evaluate the workability of conditioned marly limestone.</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

GEP-Based Prediction of Workability of Conditioned Marly Limestone in EPB Machine Chambers

  • 期刊论文

  • 发表日期：

    2025年9月23日

查看全文
  • Hamidreza Taherkhani,
  • Jafar Hassanpour,
  • Hassan Bakhshandeh Amnieh,
  • Emad Khorasani,
  • Jamal Rostami,
  • Ali Mohammad Saheb

摘要

The performance of Earth Pressure Balance (EPB) machines strongly depends on the proper conditioning of excavated materials. This research presents a novel gene expression programming (GEP) model to predict the slump value of conditioned marly limestone in the pressure chamber of EPB-TBMs. A total of 97 slump tests were conducted to examine the effects of bentonite slurry on improving the workability of crushed marly limestone. Fine particle percentage (F), bentonite slurry concentration (CB), and slurry injection ratio (SIR) were selected as input variables, while slump value (S) served as the output. The GEP model was developed using 78 training and 19 testing data sets, and its performance was compared to a linear multivariable regression (LMR) model. Evaluation of the models were conducted using four statistical metrics: coefficient of determination (R2), variance accounted for (VAF), root mean square error (RMSE), and mean absolute error (MAE). The results showed that GEP outperformed LMR, achieving R2 = 0.913, VAF = 89.2%, RMSE = 2.294, and MAE = 1.98 in the test phase. Moreover, residual analysis was performed and showed that residuals were normally distributed and randomly scattered around zero, confirming no systematic bias or overfitting. Finally, a sensitivity analysis revealed that SIR had the greatest impact on slump prediction, while F had the least. The GEP model developed in this study could be utilized as a reliable tool to more accurate prediction of the slump value to evaluate the workability of conditioned marly limestone.