New methods for multivariate analysis of power quality in large amounts of data

• Applicant: Dr.-Ing. Max Domagk
• Subject Area: Electrical Energy Systems, Power Management, Power Electronics, Electrical Machines and Drives
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 521923789

Power quality (PQ) is becoming increasingly important in the electrical power supply of modern industrial societies such as Germany. This is due, on the one hand, to the increased use of devices and systems that lead to a reduction in PQ due to their impact on e.g. voltage and current distortion within networks and, on the other hand, to the growing number of devices and systems that react increasingly sensitive to a reduction in PQ. The accelerated transition to distributed generation as well as the introduction of electromobility and electrical storage have a non-negligible impact on PQ. Therefore, there is an increasing need for comprehensive knowledge on the character and possible temporal changes of PQ. More and more PQ measuring devices are therefore being installed in the electrical power supply networks, which is why the already enormous amount of available measurement data will again increase significantly in the near future. Currently, the collected data is often only used for individual fault analysis. In the case of regular analyses, these are mostly limited to simple standard evaluations, which only provide a week-by-week qualitative statement (yes/no) about compliance with specified limit values. Most of the information about the behaviour or characteristics of PQ within the network remains unused because suitable tools and indices for their automated analysis do not exist and manual techniques are unsuitable due to the enormous amount of data. However, it is precisely this information that has a high economic value. For example, they contribute to an early detection of local or global negative developments with regard to PQ. In addition, they can help to optimize the number and distribution of metering locations in the network and to work proactively towards ensuring adequate PQ on a sustainable basis. The additional information also forms the basis for the proper further development of standards and guidelines as well as the future integration of aspects of PQ into network planning. For the reasons mentioned above, there is currently a high demand for new and effective procedures, techniques and evaluation indices for the automated analysis of large data sets from long-term measurements of PQ parameters. The future new methods will be based on multivariate analysis techniques, especially for the identification and quantification of similarities and interdependencies between quality parameters and/or measurement locations and/or evaluation periods. Similarities refer to the shape of the course or pattern of different time series and have a rather qualitative character. Interdependencies, on the other hand, have a quantitative character and refer in particular to different types of correlations.

• DFG Programme: Research Grants