Members of the Artifact project participated in the 13th Urban Drainage Modelling (UDM) Conference, held from September 15–19 in Innsbruck, Austria. During this recognized international forum, our team had the opportunity to present their research contributions.

Luka Vinokić delivered a presentation titled “Digital Twins of Urban Drainage Systems: ML-assisted Algorithm for Processing Sensor Data.”
This research addresses the challenge of data quality in Digital Twins of Urban Drainage Systems (UDS). Sensor networks often produce data with missing values and anomalies due to factors such as sensor malfunction, hardware limitations, or environmental conditions. To address this, Luka together with his co-authors, developed an advanced machine learning–based algorithm for automated anomaly detection and missing data estimation. The algorithm, which employs an ensemble of ML models, was tested on a synthetic dataset representing part of the Belgrade stormwater system.

Jasmina Moskovljević presented a poster titled “Urban Flood Prediction and Mapping Using Machine Learning and Deep Learning”.
This study provides a comprehensive literature review of recent applications of ML and DL methods for urban flood prediction and mapping. Furthermore, a Bayesian Linear Mixed Model (BLMM) analysis was conducted to evaluate the performance of different models under varying conditions. The results offered valuable insights into how key input factors influence flood predictions and flood maps.

All presented works can be found at the following link: PUBLICATIONS

Our participation in the conference in Innsbruck proved extremely valuable, offering the opportunity to present our work to the scientific community, exchange knowledge with colleagues worldwide, gain fresh perspectives, and establish collaborations that will support our future research. Looking ahead, the 14th UDM Conference will take place in Exeter, United Kingdom, in 2028, marking another important occasion for the community to come together.