Lab TP uses data to help with decision-making
A Bouygues Travaux Publics service created in 2016
Lab TP aims to design predictive models to help with the management and maintenance of tunnelling machines and improve safety and productivity when digging tunnels. With better understanding of the internal and external interactions of tunnelling machines, certain technical issues can be anticipated and resolved, increasing the rate of progress in digging. At most of our tunnelling sites, data generated by the tunnelling machines is centralised at a worksite data hub and then analysed by a data scientist and engineer. The benefits of this are more accurate tracking of productivity, real-time management of site activity and improved feedback. Raw site data is also sent automatically to the central data platform, which identifies the data and stores it securely over the long term. This data is then made available to data scientists to model and create predictive algorithms by means of machine learning, which will further improve our productivity, just like the many decision-making tools using big data already in operation at our sites.
Thesis proven!
Titled “Modelling the behaviour of tunnelling machines and impact on their environment”, the thesis of Mehdi Mahmoudysepehr, a doctoral student at the Construction 4.0 Chair in partnership with Centrale Lille engineering school, endeavours to analyse big data collected on the ground in order to better anticipate and solve technical issues. And also to use statistical learning methods (machine learning and deep learning) to model the tunnelling machine’s behaviour and optimize its control.
Further information
30
is the number of innovations planned in the Lab TP roadmap
What is machine learning?
Machine learning is an artificial intelligence technology that allows computers to learn without having to be programmed specifically for this purpose. To learn and develop, computers nevertheless need a lot of data (big data) to analyse and learn from.
Did you know?
A big data application is in use at the Eole site
It models the tunnelling machine’s behaviours and provides a terrain-machine interaction indicator to help teams control the machine. This provides a data-driven approach to make strategic decisions in real time.
What is the principle of TopoEveryWhere?
Connecting from a single base to several networks of sensors autonomously measuring the impact of tunnelling projects and displaying the measurements on dashboards. An actual control room to track the behaviour of neighbouring structures.