Short Term Solar Power Prediction Using Hybrid Two Layered Decomposition Technique Based Optimized ELM

Abstract

The rapid growth in power demand increased the per capita consumption of power. In this scenario, the nonconventional energy sources play a significant role in a power system. Solar power is one of the renewable sources RES, popularly used to meet energy demand. The increase in the PV integration into the main grid makes the solar power prediction an essential aspect as it helps in the reduction of different power quality issues and thus enhancing the system reliability. The nonlinear nature of solar power makes the prediction difficult hence a precise prediction technique is required for an accurate result. This paper proposes a hybrid technique is proposed for 5min- ahead solar power prediction. The hybrid model comprises EMD, VMD, and ELM optimized by phase angle particle swarm optimization (PA-PSO). To validate the accuracy and effectiveness of the proposed model a solar power data series is considered. 5min solar power data from New Jersey, is considered as interpretive examples for evaluating the model efficiency. The experimental result shows that the proposed model outperforms other techniques considered over the different prediction horizon.