Scholarlink

<p>Concrete, the second most-consumed construction material, has seen significant advancements in the past two decades, including High-Performance Fiber-Reinforced Concrete (UHPFRC) and Functionally Graded Concrete (FGC). HPFRC aims to improve mechanical properties, reduce weight, increase environmental resistance, and reduce material costs. However, it is expensive due to high-quality materials and complex mix design, which can lead to brittleness under tensile stress. These issues can be resolved by analyzing Functionally Graded Reinforced Concrete (FGRC) beams with High Performance Fiber-Reinforced Concrete (HPFRC) layers to improve structural performance and by using machine learning to accurately predict the mechanical properties of the beams. This study utilizes simulation to analyse the stress responses, deflections, and failure processes of a 100 × 100 × 700&#xa0;mm beam made of HPFGC with varying cement replacement percentages. The beam features outer layers of steel fiber concrete (SFC) with an 80&#xa0;mm thickness, enhancing tensile strength and ductility, while the core is composed of normal concrete (NC). The parameter data collected from simulations including elastic modulus, yield strength, Cross-sectional dimensions, and axial load. The data was pre-processed using criteria importance through inter-criteria correlation (CRITIC) normalization to ensure efficient analysis. Mechanical properties of High-Performance Functionally Graded Concrete (HPFGC) beams were then predicted using an extreme learning machine (ELM) to determine the ideal number of hidden layers and activation function using boosted dipper throated optimization (BDTO). BDTO-ELM model outperformed existing predictive models, achieving a prediction accuracy of 96.5%, with RMSE values of 0.04 and MAE values of 0.017. According to the modelling outcomes, the proposed BDTO-ELM model has an enormous amount of efficacy for predicting the mechanical properties of the HPFGC beam.</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

Cracking behavior analysis and mechanical property prediction of high-performance functionally graded concrete beams using BPTO based extreme learning machine

  • 期刊论文

  • 发表日期：

    2025年8月26日

查看全文
  • Ajay Prabhakar,
  • Shambhu Sharan Mishra

摘要

Concrete, the second most-consumed construction material, has seen significant advancements in the past two decades, including High-Performance Fiber-Reinforced Concrete (UHPFRC) and Functionally Graded Concrete (FGC). HPFRC aims to improve mechanical properties, reduce weight, increase environmental resistance, and reduce material costs. However, it is expensive due to high-quality materials and complex mix design, which can lead to brittleness under tensile stress. These issues can be resolved by analyzing Functionally Graded Reinforced Concrete (FGRC) beams with High Performance Fiber-Reinforced Concrete (HPFRC) layers to improve structural performance and by using machine learning to accurately predict the mechanical properties of the beams. This study utilizes simulation to analyse the stress responses, deflections, and failure processes of a 100 × 100 × 700 mm beam made of HPFGC with varying cement replacement percentages. The beam features outer layers of steel fiber concrete (SFC) with an 80 mm thickness, enhancing tensile strength and ductility, while the core is composed of normal concrete (NC). The parameter data collected from simulations including elastic modulus, yield strength, Cross-sectional dimensions, and axial load. The data was pre-processed using criteria importance through inter-criteria correlation (CRITIC) normalization to ensure efficient analysis. Mechanical properties of High-Performance Functionally Graded Concrete (HPFGC) beams were then predicted using an extreme learning machine (ELM) to determine the ideal number of hidden layers and activation function using boosted dipper throated optimization (BDTO). BDTO-ELM model outperformed existing predictive models, achieving a prediction accuracy of 96.5%, with RMSE values of 0.04 and MAE values of 0.017. According to the modelling outcomes, the proposed BDTO-ELM model has an enormous amount of efficacy for predicting the mechanical properties of the HPFGC beam.