Developing an explainable artificial intelligent (XAI) model for predicting pile driving vibrations in Bangkok’s subsoil
บทความในวารสาร
ผู้เขียน/บรรณาธิการ
กลุ่มสาขาการวิจัยเชิงกลยุทธ์
รายละเอียดสำหรับงานพิมพ์
รายชื่อผู้แต่ง: Youwai S.; Pamungmoon A.
ผู้เผยแพร่: Springer
ปีที่เผยแพร่ (ค.ศ.): 2025
วารสาร: Neural Computing and Applications (0941-0643)
Volume number: 37
Issue number: 18
หน้าแรก: 12881
หน้าสุดท้าย: 12902
จำนวนหน้า: 22
นอก: 0941-0643
eISSN: 1433-3058
ภาษา: English-Great Britain (EN-GB)
บทคัดย่อ
This study presents an explainable artificial intelligence model for predicting pile driving vibrations in Bangkok’s soft clay subsoil. A deep neural network was developed using 1,018 real-world pile driving measurements, encompassing various pile and hammer characteristics, sensor locations, and vibration measurement axes. The model achieved a mean absolute error of 0.276, outperforming traditional empirical methods and other machine learning approaches. SHapley Additive exPlanations analysis was employed to provide both global and local explanations of the model’s predictions. Global explanations revealed complex relationships between input features and peak particle velocity, with distance from the pile driving location emerging as the most influential factor. Local explanations enabled interpretation of individual predictions, allowing for targeted optimization of pile driving parameters. Nonlinear relationships and threshold effects were observed, providing new insights into vibration propagation in soft clay. A web-based application was developed to facilitate adoption by practicing engineers, offering both predictive capabilities and explanations for each prediction. This research contributes to geotechnical engineering by offering a more accurate, nuanced, and interpretable approach to predicting pile driving vibrations, with implications for optimizing construction practices and mitigating environmental impacts in urban areas. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.
คำสำคัญ
ไม่พบข้อมูลที่เกี่ยวข้อง
ติดต่อสอบถามเพิ่มเติมได้ที่