Quantitative research on impact of ambient temperature and module temperature on short-term photovoltaic power forecasting

Abstract

As one of the most correlative impact factors of photovoltaic (PV) power output, the PV module temperature plays very important role in PV power forecasting, but often be confused with ambient temperature. In this paper, the research on impacts of ambient temperature and PV module temperature on power output of PV modules is analyzed to explore the differences and similarities of the two kinds of temperatures. Firstly, based on Mutual Information Theory, the quantitative research is presented to study the correlations between the power output of PV modules and the two temperatures under four different weather classifications: sunny day, cloudy day, shower day, and heavy rainy day, marked by A, B, C, and D respectively. And then, three different short-term step-wise models based on support vector machine (SVM) are applied to forecast the power output for photovoltaic. Step-wise model means that there are two layers in a model, in which the first layer is to predict the impact factors while the second layer is a mapping model between PV power output and the impact factors. The results indicate that for PV power output, the impact of PV module temperature is stronger about 20%∼30% than ambient temperature, and the forecasting results measured by Mean Absolute Percentage Error (MAPE) show that the model only using PV module temperature as the inputs is more accurate.