Distributed photovoltaic power station state prediction method and system

The invention discloses a distributed photovoltaic power station state prediction method and system, and relates to the technical field of photovoltaic power station detection, and the method comprises the steps: collecting the historical operation data of a grid-connected cabinet in a station, and building a first prediction model; dynamically adjusting a time threshold according to the first prediction model and the data set; constructing a second prediction model and a third prediction model in combination with the first prediction model; integrating judgment logic according to the output of the three models, and predicting the operation state of the power station. The dynamic time threshold adjusting module is introduced, so that the system can be continuously self-optimized and calibrated, the robustness is improved, the false alarm risk caused by environment change is avoided, and the reliability is remarkably improved. A plurality of communication, power generation and tripping models and integrated ju