Non-technical loss and power blackout detection under advanced metering infrastructure using a cooperative game based inference mechanism

To efficiently detect non-technical loss and power blackout in micro-distribution systems, this study proposes using a cooperative game (CG) based inference mechanism under the advanced metering infrastructure technique. Fractional-order Sprott system is designed to extract specific features between the profiled usages and the measurement usages in real time analysis. The fractional-order dynamic errors are positive correlated with the changes in load usages, including normal conditions, electricity fraudulent events, and power blackout events. Then, multiple agents in a game and multiple CG based inference mechanisms are used to locate abnormalities in micro-distribution systems. For energy management applications, the proposed inference mechanism can identify the 2.5–20% irregular usages during normal demand operations. In addition, it can also identify the large changes >20% in usages, while a micro-distribution system is disconnected to operate in the islanded mode within a few hours. This function can address an outage occurrence and then quickly resume service using the service restoration strategy and distributed generations in a local grid. Using a medium-scale micro-distribution system, computer simulations are conducted to show the effectiveness of the proposed inference model.