Impact of Renewable Energy Sources on Artificial Neural Networks for Optimal Power Flow: A Contemporary Analysis

Artificial neural networks (ANN) have significantly improved speed and frequency of solving optimal power flow (OPF) problems. It also proved its ability to capture generation outputs and their characteristics for renewable energy sources (RES) and integrate them in OPF problems. This paper presents an evaluation of the impact of RES units on ANN accuracy when the ANN is trained with no presence of RES units in the system. It analyzes how accuracy of this ANN is affected compared to OPF solution from MATPower library; firstly when wind and solar RES units are injected as additional power sources and then when the existing conventional generation units are replaced by RES units. This work promotes for better understanding of how RES units affect ANN trained for OPF with no RES units. A case study of IEEE 30 bus system is presented along with numerical comparisons between OPF solutions from the ANN and MATPower and illustrated using 3D plots.