Scholarlink

<p>Cumulative energy dissipation is a critical parameter for assessing the level of damage to structural members in energy-based seismic design of structures. This study introduces a machine learning-aided framework to accurately and efficiently predict the cumulative energy dissipation of code-compliant reinforced concrete beams and columns by correlating with damage levels. A comprehensive dataset, comprising features and dissipated energies of 13,289 columns and 8,060 beams, serves as the foundation to train and test the models. Results indicated that the proposed method, particularly when supported by extreme gradient boosting and random forest, achieves high accuracy in predicting cumulative energy dissipation across varying damage levels, as validated by key performance indicators. Additionally, sensitivity analyses explore the impact of algorithm-specific hyperparameters on prediction accuracy, further refining the model’s reliability. Consequently, proposed framework can serve as a valuable tool for structural engineers and researchers, facilitating more accurate and efficient seismic damage assessments in frame-type structures.</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

Integrating energy-based principles to damage quantification of code compliant frame members through machine learning methods

  • 期刊论文

  • 发表日期：

    2026年6月29日

查看全文
  • Ziya Muderrisoglu,
  • Ahmet Anıl Dindar,
  • Ahmet Güllü,
  • Ali Bozer,
  • Hasan Özkaynak

摘要

Cumulative energy dissipation is a critical parameter for assessing the level of damage to structural members in energy-based seismic design of structures. This study introduces a machine learning-aided framework to accurately and efficiently predict the cumulative energy dissipation of code-compliant reinforced concrete beams and columns by correlating with damage levels. A comprehensive dataset, comprising features and dissipated energies of 13,289 columns and 8,060 beams, serves as the foundation to train and test the models. Results indicated that the proposed method, particularly when supported by extreme gradient boosting and random forest, achieves high accuracy in predicting cumulative energy dissipation across varying damage levels, as validated by key performance indicators. Additionally, sensitivity analyses explore the impact of algorithm-specific hyperparameters on prediction accuracy, further refining the model’s reliability. Consequently, proposed framework can serve as a valuable tool for structural engineers and researchers, facilitating more accurate and efficient seismic damage assessments in frame-type structures.