Scholarlink

<p>Freeze–thaw cycles severely degrade clay soils in cold regions, threatening infrastructure stability. This study utilizes cherry marble powder (CMP), an eco-friendly industrial by-product, to enhance soil resilience. Experimental results show CMP increases unconfined compressive strength by up to 97.75% and reduces mass loss by 20-35% after 15 freeze–thaw cycles, with optimal performance at 21.5% moisture and 40% CMP. Remarkably, energy absorption improves by 25%, indicating enhanced toughness. To minimize experimental burden, a Bayesian regularization-based ANN model with a single hidden layer containing 14 neurons is developed, achieving exceptional prediction accuracy (<i>R</i><sup>2</sup> = 0.9959). This work pioneers the integration of sustainable waste valorization and advanced machine learning for soil stabilization, offering a dual innovation in environmental sustainability and smart geotechnical design. The findings provide a practical, data-driven framework for resilient infrastructure in freezing climates, reducing reliance on traditional, resource-intensive methods.</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

Cherry Marble Powder’s Effect on Clay Soil Stability in Freezing Regions Based on Experimental and Advanced Artificial Neural Networks Investigation

  • 期刊论文

  • 发表日期：

    2026年5月6日

查看全文
  • Ibrahim Haruna Umar,
  • Müge Elif Fırat,
  • Orhan Atila,
  • Hang Lin

摘要

Freeze–thaw cycles severely degrade clay soils in cold regions, threatening infrastructure stability. This study utilizes cherry marble powder (CMP), an eco-friendly industrial by-product, to enhance soil resilience. Experimental results show CMP increases unconfined compressive strength by up to 97.75% and reduces mass loss by 20-35% after 15 freeze–thaw cycles, with optimal performance at 21.5% moisture and 40% CMP. Remarkably, energy absorption improves by 25%, indicating enhanced toughness. To minimize experimental burden, a Bayesian regularization-based ANN model with a single hidden layer containing 14 neurons is developed, achieving exceptional prediction accuracy (R2 = 0.9959). This work pioneers the integration of sustainable waste valorization and advanced machine learning for soil stabilization, offering a dual innovation in environmental sustainability and smart geotechnical design. The findings provide a practical, data-driven framework for resilient infrastructure in freezing climates, reducing reliance on traditional, resource-intensive methods.