Analyzing Optimal Battery Sizing in Microgrids Based on the Feature Selection and Machine Learning Approaches

Microgrids are becoming popular nowadays because they provide clean, efficient, and lowcost energy. require bulk storage capacity to use the stored energy in times of emergency or peak loads. Since microgrids future renewable energy, technology employed should be optimized power balancing. Batteries play a variety essential roles daily life. They used at hours during time emergency. There different types batteries i.e., lithium-ion batteries, lead-acid etc. Optimal battery sizing is challenging problem that limits modern technologies such as electric vehicles, Therefore, it imperative assess optimal size for particular system microgrid according its requirements. The can assessed based on features life, throughput, autonomy, In this work, mixed-integer linear programming (MILP) newly generated dataset studied computing terms autonomy. considered dataset, each instance composed 40 attributes battery. Furthermore, Support Vector Regression (SVR) model predict capability input autonomy importance SVR model. relevant selected utilizing feature selection algorithms. performance six best-performing algorithms analyzed compared. experimental results show improve proposed methodology. Ranker Search algorithm with attains highest Spearman’s rank-ordered correlation constant 0.9756, 0.9452, Kendall 0.8488, root mean squared error 0.0525.