Performance Analysis of Neural Network Training Algorithms and Support Vector Machine for Power Generation Forecast of Photovoltaic Panel

Current energy policies are encouraging the connection of power generation based on low-polluting technologies, mainly those using renewable sources to distribution networks. The photovoltaic (PV) systems have experienced a great growth around the word in last years. Hence, it becomes increasingly important to understand technical challenges, facing high penetration of PV systems at the grid, especially considering the effects of intermittence of this source on the power quality, reliability and stability of the electric distribution system. In another hand, the connections of distributed generators, by PV panels, changes voltage profile at low voltage power systems. This fact can affect the distribution networks on which they are attached causing overvoltage, undervoltage, frequency oscillations and changes in protection design. In order to predict these disturbs, because of this PV penetration, this article aims to analyze seven training algorithms used in artificial neural networks for temporal prediction of the generated active power and thus the state of the distribution network in which these microgenerators are connected and, then compare its best results with the Support Vector Machine (SVM) technique. As a result it was concluded that 3 algorithms are suitable for this type of analysis with the best performance among the seven analyzed was the Bayesian Regularization and that Artificial Neural Networks are more suitable for this problem than the SVM.