Regional distributed photovoltaic prediction method and system, terminal and medium

The invention discloses a regional distributed photovoltaic prediction method, which is characterized by comprising the following steps of: 1, acquiring environmental parameters and real-time power of a medium-voltage photovoltaic power station and acquiring real-time power of low-voltage distributed photovoltaic equipment in a region to be predicted; step 2, constructing a medium-voltage power curve of each medium-voltage photovoltaic power station and a low-voltage power curve of each low-voltage distributed photovoltaic device, and obtaining the medium-voltage power curve associated with each low-voltage power curve; 3, superposing the real-time power of the low-voltage power curve to the medium-voltage power curve associated with the low-voltage power curve to generate a superposed power curve; 4, constructing a feature sample through the environmental parameters of the medium-voltage photovoltaic power station, inputting the feature sample into the BP neural network, outputting predicted power, and compa