The application of black-box modeling in long-term tunnel deformation monitoring

Tunnel closures related to maintenance and reconstruction works can lead to large economical costsand should therefore be avoided. This paper explores the use of novelty detection algorithms forlong-term tunnel monitoring. The aim is to detect tunnel damage in an early stage, as such providinga tool to support the asset management. The proposed strategy is applied to the monitoring of theWaasland tunnel in Antwerp, where the deformations and temperatures have been monitored overa period of 14 months. The case demonstrates that novelty detection by means of principalcomponent analysis enables the identification of minor changes in the tunnel response, and cantherefore be embedded in an early detection warning system.

Publication year:2021