AN ENERGY MANAGEMENT SYSTEM USING OPTIMIZED HYBRID ARTIFICIAL NEURAL NETWORK FOR HYBRID ENERGY SYSTEM IN MICROGRID APPLICATIONS

Abstract

Different energy sources are typically incorporated into coordinated MGS (Micro Grid Systems) using energy management systems. It is challenging to integrate acceptable energy management models in MGS mainly due to the unpredictable nature, availability estimations and complexities in regulating RES (Renewable Energy Sources). Energy policies are encouraging incorporation of RES while reducing the usage of fossil-based fuels resulting in the need to optimize RES.This study's major goal is to lower running costs of grid-connected MGSs while predicting PV (photovoltaic) based electricity and load demands in near future. In order to enhance the performance of micro-grids, this work focuses on creating a technique for integrating optimized ANN (artificial neutral networks) into an EMS (Energy Management System). The schema called EMS-HANN (Energy Management System - Hybrid ANN) is proposed in this work and it includes forecasts, planning, data gathering, and HMI (human-machine interfaces) components. Day-ahead PV power and load demand estimates are combined with a 3-level SWT (stationary wavelet transforms) as part of the forecasting module's enhanced hybrid forecasting technique and GWO-HANN (grey wolf optimization-based Hybrid Artificial Neural Network). The scheduling module employs AEHO (Adaptive Elephant Herding Optimisation)-based scheduling to deliver the optimal power flow for grid-connected MGS. Subsequently, DAQ and HMI modules monitor, analyse, and change forecast and schedule input variables. The proposed model for applications of MGS is implemented along with current algorithms in MATLAB/Simulink platform where outcomes demonstrate better performances of the suggested model as compared to comparable efforts.