On Tuesday 26 March, together with the Dike Monitoring Network and TU Delft, we organised a DigiShape day on dike monitoring and AI. We looked at the state of the art of different methods of dike monitoring and discussed a number of AI systems developed together by academia and practice. A video impression can be found on LinkedIn.
Sheets and videos
Wouter Zomer – Introduction Network Dikemonitoring
Chris Karman – Introduction DigiShape
Jeroen Baars – Introduction and perspective from the water board
Jeroen Baars of Hoogheemraadschap Hollands Noorderkwartier ttok us through the perspective of the water board. In what way can Artifical Intelligence offer solutions within dike management? In what ways is AI already being deployed in the water board?
Joost Stenfert – AI Toolbox for inspection monitoring of flood defences
Joost Stenfert discusses the AI Toolbox for inspection monitoring of flood defences, which is under development at HKV line in water.
The use of AI has the potential to radically change inspection and monitoring of flood defences. Water boards inspect many thousands of kilometres of flood defences every year. To gain insight into the actual strength of flood defences, it is becoming increasingly important to look at differences in condition compared to the assumptions of testing and assessment. Such inspections are labour-intensive. Administrators indicate that recording damage images manually and processing inspections is time-consuming. This has led to the ambition to develop an AI Toolbox for flood defences together with water boards. Joost shows examples and ambitions and engages with participants about the Toolbox’s ideas and possible further developments.
Muriel Serrurier Schepper – Training AI models
Muriel Serrurier Schepper of the AI Annotation Lab will talk about training AI models. The AI Annotation Lab is a social enterprise that provides high-quality, responsible data annotation at a central location in the Netherlands. In doing so, it works with people with a distance to the labour market who, with their different backgrounds and perspectives, ensure high-quality and ethical data annotation.
Juan Pablo Aguilar Lopez – The role of sensors in AI systems for dike monitoring
Juan Pablo Aguilar Lopez from TU Delft will discuss the role of sensors in AI systems for dike monitoring. Machine Learning algorithms, developed from historical data, combined with smart and advanced sensors form the core of an AI system. Juan will demonstrate this through a number of current projects being carried out by him and his students.
Kin Sun Lam – Digitwin and proven dike strength
Kin Sun Lam shows that in a DigiTwin for flood defences, subsurface, dyke monitoring and AI/data science come together. Dike monitoring over time provides the input that data science and AI can be used to improve and sharpen dike schematics and models. This automated learning system gives the embankment manager ever better insight into its embankments under different conditions.
He will illustrate this using a pilot at Hoogheemraadschap Hollands Noorderkwartier. With gauge monitoring and data science, the barrier manager gains insight into surviving (wet) conditions and the corresponding return times. This proven dike strength may be redeemable in the form of a tightened (updated) dike failure probability.
