Prediction and influencing factors of photovoltaic system power generation of a single house based on continuously measured data

Abstract

With the development of photovoltaic (PV) power generation systems in single houses, research has recently focused on the prediction of PV power generation to match PV power generation with building energy consumption characteristics. However, prediction models for PV power generation under different weather conditions based on the actual monitoring data of the PV power generation system in a single house are still lacking. The present study is based on the actual monitoring of PV power generation data of a single house with a PV system that was installed in Beijing 5 years ago. We analysed the main weather factors affecting PV power generation and power conversion efficiency (PCE). The results showed a positive correlation between power generation and solar radiation, with the ambient temperature (Ta) and total cloud cover (TCC) being the main weather factors affecting power generation and PCE. Based on the analysis results, a support vector regression (SVR) model was established to predict PV power generation. In addition, the mean absolute error (MAE), mean absolute percentage error (MAPE) and coefficient of variation (R2) of the SVR model were 3.39, 34.7% and 0.86, respectively, which can accurately be used to predict PV power generation under different weather types.