Scholarlink

For assessing the safety performance of a structure in service, the ultimate pile bearing capacity is an important parameter. The literature indicates that experimental data on pile bearing capacity can be utilized to make quick and accurate predictions. Random forest is employed for these predictions, while five distinct algorithms are utilized to optimize the process: PSO, SSA, WSO, GWO, and DBO are all optimization algorithms used in engineering and computer science. These algorithms are designed to solve complex optimization problems by mimicking the behavior of animals in their natural habitats. An unused subset is selected randomly and evaluated with appropriate metrics to assess the model. Based on the results, the R2 values for the five models are 0.9103, 0.8766, 0.9410, 0.9133, and 0.9687 for RF-PSO, RF-SSA, RF-WSO, RF-GWO and RF-DBO. As far as Mean Absolute Percentage (MAP) is concerned, the RF-DBO model has the lowest optimal MAP. The RF-DBO prediction model has the lowest RMSE (3.6232) and MAPE among all models, making it the most accurate, stable, and effective in predicting pile capacity. It outperforms the RF-PSO, RF-WSO, and RF-GWO models in these categories.

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

Research on Pile Bearing Capacity Prediction Model Based on Optimized Random Forests

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Shunbo Li,
  • Bohang Chen,
  • Mingwei Hai,
  • Qi Zhang,
  • Zhangyu Chen,
  • Bin Zhou,
  • Yanxiu Guo,
  • Zhuo Zhao

摘要

For assessing the safety performance of a structure in service, the ultimate pile bearing capacity is an important parameter. The literature indicates that experimental data on pile bearing capacity can be utilized to make quick and accurate predictions. Random forest is employed for these predictions, while five distinct algorithms are utilized to optimize the process: PSO, SSA, WSO, GWO, and DBO are all optimization algorithms used in engineering and computer science. These algorithms are designed to solve complex optimization problems by mimicking the behavior of animals in their natural habitats. An unused subset is selected randomly and evaluated with appropriate metrics to assess the model. Based on the results, the R2 values for the five models are 0.9103, 0.8766, 0.9410, 0.9133, and 0.9687 for RF-PSO, RF-SSA, RF-WSO, RF-GWO and RF-DBO. As far as Mean Absolute Percentage (MAP) is concerned, the RF-DBO model has the lowest optimal MAP. The RF-DBO prediction model has the lowest RMSE (3.6232) and MAPE among all models, making it the most accurate, stable, and effective in predicting pile capacity. It outperforms the RF-PSO, RF-WSO, and RF-GWO models in these categories.