Two BFH researchers win competitive research funding

14.09.2022 Researchers Vidushi Bigler and Angela Meyer from the School of Engineering and Computer Science TI have each won a coveted research grant from the Swiss National Science Foundation SNSF. BFH congratulates them on this achievement.

The SNSF awards Practice-to-Science grants to experts with proven experience in practice who wish to strengthen the academic component of their dual scientific-practical skills profile. In this year’s award, two researchers from the School of Engineering and Computer Science (TI) at Bern University of Applied Sciences (BFH) prevailed in an extremely competitive environment: Vidushi Bigler and Angela Meyer impressed the jury with their respective projects and secured three years of funding for their research.

angela-vidushi

Predicting the condition of industrial fleets better

The project “Artificial Intelligence for Improving the Reliability and Resilience of Industrial Fleets” by Angela Meyer seeks to develop machine learning algorithms that help improve fault diagnosis in industrial fleets such as wind farms or photovoltaic power plants. Today, these plants are monitored around the clock using sensors to detect and diagnose operating faults at an early stage. Machine-learning models are already in use for condition monitoring, but often cannot be used with high accuracy for all fleet components. This leads to delays and errors in judgement in the detection and diagnosis of equipment malfunctions. The research project will develop and test new deep-learning approaches to overcome these drawbacks and thus improve the reliability and resilience of industrial fleets.

Modelling of the Swiss network of rivers and streams as a graph

Climate change is expected to have a major impact on Swiss rivers and lakes. For people, animals, plants and industrial applications, the availability of water is not the only crucial aspect here, but also the temperature of the water. In light of this, long-term temperature forecasts over several decades are needed, providing an important basis for decision-making at the federal level. However, accurately simulating stream temperatures over a large range of space and time using multiple climate models is challenging and requires considerable computer resources. A supercomputer, for example, would take several years to model the entire Swiss river and stream network. An innovative solution based on deep learning and graph theory is the proposition of the project “Spatio-temporal graph convolutional networks – a novel deep learning approach to forecasting river temperatures”. Vidushi Bigler from the Institute for Optimisation and Data Analysis (IODA) has no doubt that the results of the project will have a considerable practical impact on the management of water resources in Switzerland. She anticipates that the impact of the project will be significant for federal and cantonal authorities, hydropower and industrial companies, municipalities and the international scientific community.