Measurement technology is currently not capable of efficiently and fully measuring the temperatures in an electric motor – the necessary sensors and their installation are simply too expensive. Temperatures in the rotating parts are particularly difficult to measure. The problem, however, is not only the lack of measuring instruments, but also the deviations that can occur during mass production. That is why manufacturers must provide for additional safety reserves, which in turn reduce the efficiency of the motors. As part of a project funded by the German Research Foundation (DFG), scientists from the Department of Power Electronics and Electrical Drive Technology at the University of Paderborn now aim to develop software capable of estimating the temperatures at certain points. In doing so, they are paying special attention to the sensitive and expensive permanent magnets.
The researchers are looking for the solution in data-driven approaches. They are making use of artificial intelligence and machine learning to find new models for estimating the temperature in drives and other power engineering applications. This involves training their software with black-box approaches and experimental test-bench measurements in order to obtain the most precise temperature estimates possible.