AI monitoring brings new precision to infrastructure planning
For a project run by Innovation Zurich's Artificial Intelligence (AI) Innovation Sandbox, Swiss railway company Südostbahn and irmos technologies used sensors to collect information about an old bridge. This demonstrates the huge potential of data-based infrastructure monitoring.
Innovation Zurich has published the latest report from its AI Innovation Sandbox. Under the title “Sensor-based bridge monitoring – data collection and predictions for infrastructure maintenance”, Innovation Zurich, a joint initiative of the Canton of Zurich Economic Development Agency, Greater Zurich Area and Switzerland Innovation Park Zurich, investigated what data-based monitoring can achieve in terms of risk assessment and modernisation strategies for infrastructure structures. Data-based monitoring approaches are not yet an integral part of established standard-based condition assessment, which is usually extremely conservative, according to the report.
In a pilot project, Swiss South-Eastern Railwayfrom St. Gallen and irmos technologies AG, based in Technopark Zurich, examined the Reidholz Bridge on the Wädenswil-Einsiedeln line as an example. It was built in 1910 and rebuilt in 1953. The railway company wanted to clarify whether extending its use from ten to 20 years would be technically possible and economically viable. It also examined whether the methodology could be transferred to comparable bridges in its portfolio.
Using easy-to-install sensors from irmos technologies, which are specially tailored to the requirements of bridge monitoring, the research partners collected data from real-world operations during a test phase lasting several months. Cameras were also used to document train types, speeds and compositions and to validate the modelled traffic loads.
The results show that even short measurement periods enable robust statements to be made about fatigue and stiffness, facilitating the trade-off between extending the service life and replacing the structure. This is because “based on sensor data, the system supports infrastructure operators in identifying risks at an early stage and planning maintenance measures in a targeted manner,” summarises irmos founder and CEO Dr Panagiotis Martakis.
The report concludes that it is crucial that data collection is not an end in itself and that data-based methods solve a specific operational problem, such as the question: How long will the bridge actually last compared to standard-based estimates? The method delivers the greatest added value when it goes beyond individual cases: monitoring entire bridge portfolios enables comparisons, prioritisation and long-term investment strategies. To achieve this, infrastructure operators need to build up their expertise.