Abstract:

In order to solve the problem of large errors in the power generation of solar photovoltaic system under different conditions, a new method for power generation prediction of solar photovoltaic system was proposed. By analyzing the structure of solar photovoltaic power generation system, the influencing factors of solar photovoltaic power generation system were studied. Seasons and weather types were used as historical samples to select sample sources, and similar data points were searched in the historical database for predicting daily time-sharing meteorological data provided by meteorological departments as historical samples. The off-line parameter optimization data set was constructed with historical samples, and the generation power prediction model of power generation system was constructed with the kernel function limit learning machine algorithm, and the model parameters were optimized by the particle swarm optimization algorithm. The experimental results show that the mean absolute percent errors of the proposed method are 1.47% and 6.39% respectively under different conditions, and the relative variation of the power prediction errors of solar photovoltaic modules is less than 1% under comprehensive abnormal conditions. It is proved that the proposed method meets the actual prediction requirements.

Key words: photovoltaic, power forecasting, particle swarm optimization, kernel function limit learning machine