Geotechnical Condition Inference Based on Operational Data of Earth Pressure Balance Tunnel Boring Machine (EPB TBM) Using Machine Learning: A Case Study of Jakarta Mass Rapid Transit (MRT) CP104 & CP105
| Student | Emilly Puan Nastiti |
|---|---|
| Supervisor(s) | Dr. Dayu Apoji, S.T., M.T., M.Sc., Ph.D. |
| Year | 2026 |
Brief Abstract
Background
Urban tunneling projects such as MRT Jakarta face significant geological uncertainty due to the discrete and limited conventional geotechnical investigation data, which cannot fully represent continuous soil condition variations along tunnel alignments. During excavation, Earth Pressure Balance Tunnel Boring Machines (EPB TBM) continuously record operational parameters that inherently reflect the soil conditions encountered, presenting an opportunity for data-driven geotechnical inference.
Objective
This study aims to develop and evaluate machine learning models capable of inferring subsurface geotechnical conditions along the MRT Jakarta Phase 1 tunnel alignment (CP104 & CP105) using EPB TBM operational data.
Methodology
Random Forest, XGBoost, and OLS Multiple Regression are trained on six TBM operational parameters integrated with borehole and laboratory data, using sequential incremental splitting to simulate real excavation progression.
Key Contribution
This research provides a continuous geotechnical inference framework from TBM operational data, addressing spatial discontinuity limitations of conventional site investigation in urban tunneling.