Supervised and unsupervised geologic interpretation during EPBM tunneling

Uncertainty and variability of geologic conditions along a tunnel alignment strongly affect the tunneling performance and risk. To safely achieve the expected performance, operators of a tunnel boring machine (TBM) have to be aware of the changing geologic conditions. A geologic map can be helpful. However, it can only provide a general guidance since it is interpreted from limited boreholes at discrete locations. Therefore, during tunneling, the operators have to continuously interpret the geologic conditions based on signals produced by TBM sensors.  

Developing more systematic methods to interpret geologic conditions based on TBM data remains an open research question. This study implements supervised and unsupervised learning algorithms to interpret the encountered geologic conditions using continuous earth pressure balance shield machine (EPBM) operation data. This study also investigates the role of EPBM features in the learning models by evaluating the feature importance measures.


  • Apoji, D., Fujita, Y., Soga, K., “Soil Classification and Feature Importance of EPBM Data using Random Forests”, submitted to Geo-Congress 2022.
  • Apoji, D., Fujita, Y., Soga, K., 2019, “Toward Autonomous TBM: Real-Time Classification of Geologic Material during Tunneling using PCA”, UC Berkeley Geosystems Engineering Annual Research Symposium (poster).