Algorithms as decision support in tunneling

Despite all accurate planning, the real underground conditions in tunnel construction can never be predicted with certainty. Operators of tunnel boring machines (TBMs) must therefore make many complex decisions while drilling is in progress, without however compromising safety or timing.

For efficient and safe control under constantly changing conditions, operators of TBMS require extensive operating experience that is difficult to formalize and consequently to transfer to novices. However, AI algorithms can support this experience transfer, as Professor Fink's research group demonstrated in a recently published paper. Her team has developed an intelligent decision support system that can assist less experienced operators.

To do so, the AI algorithm learns from past projects how the controls in various geological conditions impacted the boring efficiency. Then, given the control parameters set by an operator and a rough knowledge of the geological conditions, it can recommend incremental control changes toward better efficiency, leading the operator step-by-step toward a better set of control parameters. The algorithm also always indicates how confident it is in its own recommendation.

The proposed framework for such a human-algorithm collaboration was developed using a real microtunnel project. Herrenknecht AG provided the data for the project. Furthermore, the D-BAUG team collaborated with the Herbert Einstein Research Groupvertical_align_bottom at the Massachusetts Institute of Technology.