Scholarlink

<p>This contribution provides a new energy consumption predictive model for producing high-performance concrete (HPC) using state-of-the-art machine learning techniques. Three of the most popular state-of-the-art regression models, Light Gradient Boosting Regression (LGBR), Extreme Gradient Boosting Regression (XGBR), and CatBoost Regression (CATR), are used in modeling energy demand. A novel hyperparameter optimization method called the Drawer Algorithm (DA) is introduced and used for improved model accuracy and generality in the training process. Beyond prediction, complete sensitivity analysis is performed to uncover and quantify the influence of the most vital features on energy consumption during different HPC production phases. The combination of robust ML models, an efficient optimization method, and complete and dense data on all of the main production phases of HPC enables the introduced framework to have robust accuracy, with the XGBR model optimized using DA (XGDA) yielding an R<sup>2</sup> of 0.9875 and an RMSE of 53.6594. This study combines sophisticated machine learning and optimization strategies with exhaustive data from every phase of high-performance concrete production to produce a strong model for forecasting and optimizing energy consumption, minimizing costs, and encouraging building sustainability.</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

Building Automated Computational Models for Predicting Energy Consumption in High-Performance Concrete Production

  • 期刊论文

  • 发表日期：

    2025年5月16日

查看全文
  • Anupam Yadav,
  • A. K. Dasarathy,
  • Rishabh Thakur,
  • Mohammed Rauf Abdulla,
  • Marwea Al-hedrewy,
  • R. Padmapriya,
  • Navin Kedia,
  • Priyadarshi Das,
  • Kamred Udham Singh

摘要

This contribution provides a new energy consumption predictive model for producing high-performance concrete (HPC) using state-of-the-art machine learning techniques. Three of the most popular state-of-the-art regression models, Light Gradient Boosting Regression (LGBR), Extreme Gradient Boosting Regression (XGBR), and CatBoost Regression (CATR), are used in modeling energy demand. A novel hyperparameter optimization method called the Drawer Algorithm (DA) is introduced and used for improved model accuracy and generality in the training process. Beyond prediction, complete sensitivity analysis is performed to uncover and quantify the influence of the most vital features on energy consumption during different HPC production phases. The combination of robust ML models, an efficient optimization method, and complete and dense data on all of the main production phases of HPC enables the introduced framework to have robust accuracy, with the XGBR model optimized using DA (XGDA) yielding an R2 of 0.9875 and an RMSE of 53.6594. This study combines sophisticated machine learning and optimization strategies with exhaustive data from every phase of high-performance concrete production to produce a strong model for forecasting and optimizing energy consumption, minimizing costs, and encouraging building sustainability.