Regresar

Energy management model for a standalone hybrid microgrid through a particle Swarm optimization and artificial neural networks approach

Abstract:

Energy management systems are usually used to integrate different energy sources into a coordinated microgrid system. However, given the variability of renewable sources and the complexity of calculating renewable resource availability and managing energy, it is not easy to incorporate efficient energy management models in a microgrid. This work focuses on developing a methodology to incorporate optimized artificial networks into a self-adaptable energy management system to improve microgrids performance. The proposed model consists of a set of artificial neural networks organized into a cascade configuration. A Particle Swarm Optimization algorithm optimizes each artificial neural network; the proposed model aims to estimate and provide information to the energy management system. The model is implemented in MATLAB/Simulink environment and fed with experimental data. Correlation analysis of system variables between the different artificial neural networks is performed to validate the proposed model. Simulated tests are performed with scenarios using experimental data, and an analysis of the system's response is performed in terms of the root mean squared error and linear regression. The results showed that, compared to related works, the proposed model reduced errors by 59% and 56% for single and multiple-step pbkp_rediction of energy parameter estimators. Regarding the fitness of the power estimator from the EMM for the test scenarios, an 0.1245 RMSE was obtained.