Golden jackal optimization-based regression analysis application on volume expansion estimation of cement pastes with MgO expansive additive

期刊论文
发表日期：

2024年11月5日

查看全文

Yuqing Tian,
Lina Zhang,
Guozhi Wang

摘要

The implementation of massive amounts of fly ash ( $FA$ $FA$ ) and $MgO$ $MgO$ expansive additive ( $MEA$ $MEA$ ) is impeded by their limited hydration capacity and the challenges associated with regulating delayed expansion. The current collection of literature lacks sufficient study on the application of learning methods to estimate volume expansion (V_e) of cement paste, specifically when using $FA$ $FA$ and $MEA$ $MEA$ . The process of designing and confirming methods for the evaluation of V_e involves using a dataset including 170 experimental outcomes obtained from papers. To accomplish this goal, researchers have developed support vector regression ( $SVR$ $SVR$ ). In the present research, the optimizers known as the Equilibrium Optimization Algorithm ( $EOA$ $EOA$ ), Golden jackal optimization algorithms ( $GJOA$ $GJOA$ ), and Henry gas solubility optimization ( $HGSO$ $HGSO$ ) were chosen for $SVR$ $SVR$ hyperparameters’ tuning. The $SV{R}_{HG}$ $S V R_{HG}$ , $SV{R}_{EO}$ $S V R_{EO}$ and $SV{R}_{GJ}$ $S V R_{GJ}$ models possess a substantial capacity to precisely forecast the ${V}_{e}$ $V_{e}$ of cement paste containing $MAE$ $MAE$ and $FA$ $FA$ . For $OBJ$ $OBJ$ index, the smaller value was attributed to $SV{R}_{GJ}$ $S V R_{GJ}$ , accounting for 0.0042, almost 50% smaller than $SV{R}_{EO}$ $S V R_{EO}$ at 0.0085 and roughly three times slither than $SV{R}_{HG}$ $S V R_{HG}$ by 0.0131. The superior performance of the $GJOA$ $GJOA$ -optimized $SVR$ $SVR$ model suggests that $GJOA$ $GJOA$ is particularly effective in handling the non-linear relationships present in the dataset. This finding has important implications for the design of cement pastes, particularly in optimizing the use of $MEA$ $MEA$ and $FA$ $FA$ .