4.6 Article

A hybrid stacking framework optimized method for TBM performance prediction

出版社

SPRINGER HEIDELBERG
DOI: 10.1007/s10064-022-03047-6

关键词

Tunnel boring machine; Penetration rate; Whale optimization algorithm; Stacking framework

向作者/读者索取更多资源

A new hybrid intelligent model named Stacking-WOA was developed to predict TBM penetration rate (PR) using the whale optimization algorithm. The results showed that the Stacking-WOA model outperformed the single algorithm models in predicting TBM PR and had stronger learning and generalization ability for a small number of samples.
The tunnel boring machine (TBM) performance directly affects the construction progress. The existing methods are difficult to improve and optimize the TBM performance. A new hybrid intelligent model named stacking framework optimized by whale optimization algorithm (Stacking-WOA) was developed to predict TBM penetration rate (PR) to overcome this problem. For this purpose, taking the tunnel of Shenzhen Metro Line 8 as an example, the most relevant parameters of TBM performance were collected and measured. In addition, three single models, including light gradient boosting machine, random forests, and extreme learning machine, were proposed to predict TBM PR and compared with the Stacking-WOA model. Moreover, to compare the results of these models and identify the best predictive models, several performance indices, such as the correlation coefficient (R), root mean square error, mean absolute percentage error, and a10-index, were computed to identify the best predictive models. The results show that the Stacking-WOA model is the most accurate in predicting TBM PR, and the stacking model outperforms the single algorithm model and has stronger learning and generalization ability for a small number of samples. Therefore, the stacking WOA model can accurately predict TBM PR and help guide and optimize tunneling construction projects in similar geological conditions.

作者

我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。

评论

主要评分

4.6
评分不足

次要评分

新颖性
-
重要性
-
科学严谨性
-
评价这篇论文

推荐

暂无数据
暂无数据