Short Term Hybrid PV/Wind Power Forecasting for Smart Grid Application using Feedforward Neural Network (FNN) Trained by a Novel Atomic Orbital Search (AOS) Optimization Algorithm

The un-predictable behavior of renewable energy sources due to their intermittent nature renders them very difficult to forecast the generated power. The photovoltaic and wind systems area a significant part of current working power systems. In this paper, a feed forward neural network (FNN) trained by atomic orbital search (AOS) optimization algorithms based technique is presented for the short term power forecasting of hybrid PV/Wind energy systems. The proposed technique is then compared with a feed forward neural network trained with grey wolf optimizer (GWO-NN), Barnacle mating optimizer (BMO-FNN) and whale optimization algorithm (WOA-FNN). The proposed technique effectively trains the feed-forward neural network and achieves less testing error, training error and relative error and also takes less time as compared to in-comparison techniques. AOS-FNN have capability to effectively, predict the power of hybrid PV/Wind Energy System under varying environmental Conditions.