The energy2050 conference hosted the award ceremony of the Innovation Challenge organized by Verbund. A team led by the CIMNE researcher Fernando Salazar was the winner for the Challenge Number 1.
VERBUND's dams and retaining walls are some of the most thoroughly monitored constructions in Austria. Highly sensitive measuring equipment discerns even the smallest of changes and communicates the measurement values in real time. In addition, there are regular inspections by trained personnel who carry out visual checks and further measurements, and record and evaluate the measurement data. The question in Challenge Number 1 was whether this could be supplemented by machine learning, linking the multitude of data in such a way that the performance of power plant dams could be analysed and predicted.
Fernando Salazar receiving the prize
The solution proposed by the Spanish research centre CIMNE was selected as the winner. "Effective dam monitoring systems provide a multitude of reliable technical data. While conventional statistical methods analyse these data, machine learning tools have the additional ability to interpret the data and to predict future dam performance", states Salazar. "The ultimate goal in practice is to recognise potential anomalies so early on that failures do not even have a chance to occur."