Buy article PDF
The purchased file will be sent to you
via email after the payment is completed.
US$ 35
|
Computers and Concrete Volume 34, Number 6, December 2024 , pages 773-789 DOI: https://doi.org/10.12989/cac.2024.34.6.773 |
|
|
|
Application of the tuned prediction algorithms on recycled powder mortar from construction and demolition debris |
||
Qiao Yuan, Liu Yang, Zhou Yang, Wang Runlong and Liu Tiegang
|
||
| Abstract | ||
| Recycled powder (RP) has emerged as a promising and viable alternative to traditional cementitious materials for use in concrete. The compressive strength (fC) of RP mortar has a considerable impact on the mechanical properties of RP concrete. Utilizing machine learning approaches to engineering problems, particularly when estimating the mechanical properties of construction materials, results in outstanding accuracy in forecasting and minimal experimental costs. This study aimed to provide some integrated machine-learning techniques for estimating the fC of recycled powder mortar (RPM). Initially, relevant literature is consulted to acquire data on the fC test results of 204 groups of mortars. Subsequently, the Extreme Gradient Boosting (XGB), Random Forest (RF), and Support Vector Regression (SVR) methodologies are used, followed by the optimization of their respective hyperparameters using the Chimp optimization algorithm (ChOA), in order to construct powerful forecasting approaches (XGBChOA, RFChOA and SVRChOA). According to the results, all three models exhibit excellent results in correctly anticipating the fC. By leveraging these advanced machines learning techniques and optimizing them with ChOA, the authors intended to achieve high accuracy in their predictions, thereby reducing the need for extensive experimental testing and minimizing costs associated with traditional methods of estimating the mechanical properties of construction materials. By accurately predicting the fC of RPM, these models can significantly reduce the need for extensive physical testing, leading to cost savings in material research and development. While the study mentions the generalization ability of the models, it would be beneficial to assess their performance on independent datasets or in real-world applications to confirm their practical utility. Including external factors or environmental conditions factors in the analysis could enhance the model's accuracy and robustness. The SI results of the models are roughly identical, with a small superiority of XGBChOA (SITrain=0.0334 and SITest=0.0567) compared to RFChOA, but remarkably better than SVRChOA. As it was clear from OBJ that the lowest value of OBJ belonged to XGBChOA at 1.2597, followed by RFChOA by 1.6792, and then SVRChOA equals to 1.3769. | ||
| Key Words | ||
| chimp algorithm; extreme gradient boosting; recycled powder mortar; sensitivity analysis; strength | ||
| Address | ||
| The Ninth Engineering Co., Ltd. of China First Highway Engineering Company of China Communications Construction Company (CCCC), Hangzhou City, Zhejiang Province, 311421, China | ||