The increasing expansion of renewable energies is placing on the power grids a greater burden than its developers once had in mind. This applies in particular to the dynamic fluctuations inherent with renewable energies. For the detection of critical network dynamics, conventional measurement technology is no longer sufficient and is increasingly being supplemented by high-precision, time-synchronized phase measurement units (PMUs). These enable real-time monitoring of relevant parameters such as frequency, voltage or phase angle with up to 50 samples per second. Thus, several GBytes of data per day. can be generated easily.
The first step of the researchers was to collect and process these mass data efficiently. For this purpose, the Fraunhofer IOSB-AST developed compression methods that can reduce the data volume by up to 80%. In addition, the compression helps to accelerate the subsequent data analysis.
In a second step, the PMU measured values were used to detect anomalies and identify certain malfunctions in real time. This involves the use of methods from the field of artificial intelligence to automatically evaluate the measured values. "We had to be able to automatically capture, compress and evaluate up to 4.3 million data records per day. Our approaches to error detection, for which we use AI-based methods in the project, are correspondingly complex," reports André Kummerow, a researcher at Fraunhofer IOSB-AST.