Modelling an Autonomous PV System Using Machine Learning: Study of the Contribution of Different Environmental Parameters on Power

Predicting the power of an autonomous PV system is of great interest for load management and storage. In this study, Machine Learning algorithms are used to predict the power of a PV array. In order to achieve this prediction, algorithms such as Gradient Boosting (GB), Random Forest (RF) and Bayesian Ridge (BR) are used. The aim of this paper is to study the contribution of solar irradiation, humidity and temperature under real operating conditions on performance prediction. The SHAP method implemented in Python is used for this purpose. The results show that GB has the best performance, with RMSE, MSE and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^2$</tex> values of 0.71, 0.23 and 0.993 respectively. The results confirm that sunshine and temperature contribute more to performance. In addition, solar irradiation and PV power are linear and monotonic with a correlation coefficient (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\boldsymbol{r}=0.96$</tex>) and a Spearman coefficient (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\boldsymbol{r}_{\boldsymbol{s}} =0,892$</tex>). Ambient temperature is weakly correlated with PV power, with <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\boldsymbol{r}=-0.267$</tex> and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\boldsymbol{r}_{\boldsymbol{s}}=-0.144$</tex>