The invention discloses a method for correcting super-short-term prediction of the photovoltaic power of an ARMA module in real time through a light metering network. The method includes the steps of inputting data to obtain auto-regression sliding average model parameters, inputting light resource monitoring system data and running monitoring system data, correcting the startup capacity in real time according to the running monitoring system data, setting up an auto-regression sliding average model so as to obtain a photovoltaic power super-short-term prediction result, and introducing real-time light metering station data so as to correct the photovoltaic power super-short-term prediction result in real time. By introducing the real-time light metering station data to correct the photovoltaic power super-short-term prediction result in real time, the defect that in the existing ARMA technology, photovoltaic generation power super-short-term prediction accuracy is low is overcome, and high-precision photovoltaic power generation super-short-term prediction is achieved.