While operating in three different modes with bi-directional power flow, electricity theft detection, and localization in a grid-tied microgrid (MG) become challenging tasks, particularly when the occurrence rate of the theft and its frequency patterns are random. To address this problem, a stochastic Petri net formalism is used in this paper to detect and localize the occurrence of theft in grid-tied MGs. The disturbance at any instance of time in the form of resistance above a threshold in the accumulated data of a smart meter, irrespective of the mode of operation, triggers a transition assigned to the arc, which alerts the transmission module. The affected transition and suspected user information are forwarded to the meter data management system (MDMS) for localization of the theft. A smart meter forwards its readings to the MDMS only when a theft is detected; otherwise readings are sent outside an MG once or twice a day depending on utility regulations. This low sampling rate significantly improves the overall privacy of customers. For line resistance measurement in the MG, singular value decomposition is used. In experiments, it calculates the technical and non-technical losses with 100% accuracy without knowing the exact topology of the power distribution network.
All Science Journal Classification (ASJC) codes
- Computer Science(all)
- Theft detection
- smart grids
- stochastic Petri net