|Title||Development of virtual sensors for photovoltaic inverters|
Development of virtual sensors for photovoltaic inverters.
Master Thesis in Energy Informatics
The proper integration of large-scale photovoltaic plants into the grid is a substantial part of the energy transition. The sudden failure of a high-power inverter usually causes significant disruptions in the connected distribution grid, e.g., due to frequency disturbances.
Therefore, condition monitoring using measurement values is one of the corner stones for successful plant operation.
However, sensors do not only bring positive effects. Each additional sensor brings an additional safety and failure potential, as well as increased complexity and in some cases substantial costs due to acquisition, integration and maintenance. Furthermore, many values are simply not measurable, e.g., of space limitations or because the values cannot be measured physically (e.g. system states). Virtual sensors (or soft sensors), unlike physical sensor, exist only in digital/virtual form. Deployed on a processing unit they are used to estimate/calculate the desired measurement values using either analytical, numerical, or empirical methods.
|Home institute||Department of Computing Science|
|Art der Arbeit||praktisch / anwendungsbezogen|
|Author||Jelke Wibbeke, M. Sc.|
The goal of the thesis is to improve the monitoring large-scale inverters by enabling the use of virtual sensors. For this, a sensor model has to be generated using machine learning approaches. The model shall be able to use the measurement data of physical sensors do calculate additional values and states, helpful for inverter monitoring.
Project Homepage: https://www.offis.de/en/offis/project/voraus-pv.html
 S. Kabadayi, A. Pridgen, and C. Julien, “Virtual Sensors: Abstracting Data from Physical Sensors,” 2006 Int. Symp. World Wirel. Mob. Multimed. Netw., p. 6, 2006, doi: 10.1109/WOWMOM.2006.115.
 L. Sánchez, I. Couso, J. Otero, Y. Echevarría, and D. Anseán, “A Model-Based Virtual Sensor for Condition Monitoring of Li-Ion Batteries in Cyber-Physical Vehicle Systems,” J. Sens., vol. 2017, pp. 1–12, 2017, doi: 10.1155/2017/9643279.
 Lichuan Liu, S. M. Kuo, and M. Zhou, “Virtual sensing techniques and their applications,” in 2009 International Conference on Networking, Sensing and Control, Okayama, Japan, Mar. 2009, pp. 31–36, doi: 10.1109/ICNSC.2009.4919241.
The thesis aims at students of computing science, physics, or comparable courses of study who are interested in dealing with machine learning in the field of energy informatics. It is recommended to have previous knowledge in the area of photovoltaics. Knowledge in pattern recognition and Python programming is desirable, but not mandatory. The applicant should be willing to deal with the new topic in a motivated and independent way.