Artificial Neural Network Based Particle Swarm Optimization for Microgrid Optimal Energy Scheduling

Abstract

This letter proposes an enhancement for artificial neural network (ANN) using particle swarm optimization (PSO) to manage renewable energy resources (RESs) in a virtual power plant (VPP) system. This letter highlights the comparison of the ANN-based binary particle swarm optimization (BPSO) algorithm with the original BPSO algorithm. The comparison has been made upon searching the optimal value of the number of nodes in the hidden layers and the learning rate. These parameter values are used in ANN training for microgrid (MG) optimal energy scheduling. The proposed approach has been tested in the VPP system covering MGs involving RESs to minimize the power and giving priority to sustainable resources to participate instead of buying power from the utility grid. This model is tested using real load demand recorded for 24 h in Perlis state, the northern part of Malaysia. Besides, real weather condition data are recorded by Tenaga Nasional Berhad Research solar energy meteorology for a 1-h average (e.g., solar irradiation, wind speed, battery status data, and fuel level). The results show that ANN-PSO gives precise decision compared with BPSO algorithm, which in turn prove that the enhancement for the neural net reaches the optimum level of energy scheduling.