Performance Evaluation of ANN and Fast Convergence MPPT Method for Solar Power

The conversion chain is the main component of the photovoltaic system, with the algorithms used being crucial to optimize this conversion chain. This article presents a study on the optimization of the energy produced by the photovoltaic panel, comparing the use of artificial intelligence, specifically artificial neural networks (ANN), with the fast convergence approach (FCA) developed in our previous work. The use of artificial neural networks demonstrates their efficiency and robustness in tracking the maximum power point under varying irradiation and temperature conditions. However, the results show that the fast convergence method is more accurate in various scenarios. Under variable temperature and constant irradiation conditions, this method has a Mean Absolute Percentage Error (MAPE) of only 1.57%, compared to 3.36% for the ANN. This means that the fast convergence method better tracks the maximum power point, even when the temperature fluctuates. Similarly, under constant temperature and variable irradiation conditions, the fast convergence method has a MAPE of 1.98%, still lower than the ANN's 2.09%. These results suggest that the FCA method offers better robustness and accuracy, particularly in the face of temperature variations.