Scholarlink

<p>Growing demand for sustainable construction materials has increased interest in using Rice Husk Ash (RHA) as a supplementary cementitious material to enhance concrete performance. However, traditional testing is costly and labor-intensive. This study proposes a predictive model using a Hybrid Neuro-Genetic Algorithm (ANN-GA) to efficiently estimate the compressive strength of RHA concrete. A comprehensive database of 1,192 datasets was utilized, encompassing seven key parameters: water, cement, RHA, fine aggregates, coarse aggregates, superplasticizer, and curing age. The ANN-GA model, configured with a single hidden layer of 36 neurons, demonstrated the best performance metrics. Comparative analysis against existing models showed the ANN-GA’s superiority, yielding the highest <i>R</i> value of 0.988, the lowest MAE of the 3.065, lowest RMSE of 3.883, and the least PE of 2.772%. Additionally, a parametric analysis was conducted to identify the effect of each input parameter. The findings confirm the capability of the ANN-GA model as a reliable and efficient alternative to traditional methods for predicting the compressive strength of RHA concrete, while also providing insights into optimal mix design strategies for enhanced performance.</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

Parametric Analysis and Machine Learning Model for Compressive Strength of Rice Husk Ash Concrete

  • 期刊论文

  • 发表日期：

    2025年12月11日

查看全文
  • Nolan Concha,
  • Juliana Mae Bayhonan,
  • James Paano,
  • Ashley Drykz Sultan,
  • Maria Sheiconne Aebrielle Tunay

摘要

Growing demand for sustainable construction materials has increased interest in using Rice Husk Ash (RHA) as a supplementary cementitious material to enhance concrete performance. However, traditional testing is costly and labor-intensive. This study proposes a predictive model using a Hybrid Neuro-Genetic Algorithm (ANN-GA) to efficiently estimate the compressive strength of RHA concrete. A comprehensive database of 1,192 datasets was utilized, encompassing seven key parameters: water, cement, RHA, fine aggregates, coarse aggregates, superplasticizer, and curing age. The ANN-GA model, configured with a single hidden layer of 36 neurons, demonstrated the best performance metrics. Comparative analysis against existing models showed the ANN-GA’s superiority, yielding the highest R value of 0.988, the lowest MAE of the 3.065, lowest RMSE of 3.883, and the least PE of 2.772%. Additionally, a parametric analysis was conducted to identify the effect of each input parameter. The findings confirm the capability of the ANN-GA model as a reliable and efficient alternative to traditional methods for predicting the compressive strength of RHA concrete, while also providing insights into optimal mix design strategies for enhanced performance.