A rising amount of today’s distribution grids are equipped with smart grid systems to face the problems arising from the increasing use of decentralized and renewable energy sources. In comparison to conventional reinforced grids smart grid systems need to be maintained to stay up to date. Especially the installed photovoltaic (PV) power is a very important parameter for the system. Self-learning smart grid systems would reduce the maintenance efforts. A huge step towards this is a self-detection of new PV power plants in the grid. In this paper three methods to detect unknown PV plants in distribution grids are introduced. They are tested and validated in a use case. Additionally the influence of undetected PV power plants to the accuracy of the grid state identification is considered. Because of the huge impact of different factors only rudimentary results are presented and further investigations are focused.