Scholarlink

<p>The highly nonlinear effect of nano-SiO<sub>2</sub> (NS) and steel-polyvinyl alcohol (PVA) hybrid fibers on the workability of self-compacting geopolymer concrete (SCGC) renders traditional linear regression models insufficient for precise prediction. To address this challenge, this study establishes a prediction model based on a backpropagation neural network (BPNN) with the initial weights and thresholds optimized by a hybrid genetic algorithm–ant colony optimization procedure to predict the workability of SCGC, which is denoted as the GA-ACO-BP model. Specifically, genetic algorithm (GA) was first employed for global coarse exploration to locate promising parameter regions, followed by ant colony optimization (ACO) for local fine-tuning to prevent the BPNN from being trapped in local optima. During the prediction process, NS content, steel fiber content, and PVA fiber content are selected as inputs, and flow rate, slump flow, and sieve segregation index are selected as outputs. Comparative analysis against standard models show that the GA-ACO-BP model achieves superior prediction accuracy and generalization. To be precise, the GA-ACO-BP model attained an <i>R</i><sup>2</sup> of 0.98 for slump flow prediction and an <i>R</i><sup>2</sup> of 0.97 for both prediction on sieve segregation index and flow rate, and also the root mean square error and mean absolute percentage error were substantially lower than those of the comparative models. It can be demonstrated that GA-ACO-BP is a robust computational tool for SCGC mix-design optimization and practical engineering implementation.</p> Graphical Abstract <p></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

Prediction of Workability of Nano-SiO2 and Steel-Polyvinyl Alcohol Hybrid Fiber-Reinforced Self-Compacting Geopolymer Concrete Based on the Back Propagation Neural Network Optimized by the Genetic Algorithm–Ant Colony Optimization Hybrid Algorithm

  • 期刊论文

  • 发表日期：

    2026年6月22日

查看全文
  • Puyuan Cao,
  • Peng Zhang,
  • Yi Ding,
  • Zhen Gao,
  • Canhua Lai

摘要

The highly nonlinear effect of nano-SiO2 (NS) and steel-polyvinyl alcohol (PVA) hybrid fibers on the workability of self-compacting geopolymer concrete (SCGC) renders traditional linear regression models insufficient for precise prediction. To address this challenge, this study establishes a prediction model based on a backpropagation neural network (BPNN) with the initial weights and thresholds optimized by a hybrid genetic algorithm–ant colony optimization procedure to predict the workability of SCGC, which is denoted as the GA-ACO-BP model. Specifically, genetic algorithm (GA) was first employed for global coarse exploration to locate promising parameter regions, followed by ant colony optimization (ACO) for local fine-tuning to prevent the BPNN from being trapped in local optima. During the prediction process, NS content, steel fiber content, and PVA fiber content are selected as inputs, and flow rate, slump flow, and sieve segregation index are selected as outputs. Comparative analysis against standard models show that the GA-ACO-BP model achieves superior prediction accuracy and generalization. To be precise, the GA-ACO-BP model attained an R2 of 0.98 for slump flow prediction and an R2 of 0.97 for both prediction on sieve segregation index and flow rate, and also the root mean square error and mean absolute percentage error were substantially lower than those of the comparative models. It can be demonstrated that GA-ACO-BP is a robust computational tool for SCGC mix-design optimization and practical engineering implementation.

Graphical Abstract