Energy management in PV,Wind,Battery, Fuel Cell & Supercapacitor Based DC Microgrids using Hybrid ANN-Ensemble Tree method

The dataset used in this study is developed to represent the operating behavior of a hybrid PV–wind–battery–fuel cell energy management system. It consists of four input variables: PV current (Ipv)(I_{pv})(Ipv), wind current (Iw)(I_w)(Iw), load current (IL)(I_L)(IL), and battery state of charge (SOCb)(SOC_b)(SOCb), which indicate the renewable source contribution, load demand, and battery energy condition. The total renewable current is calculated as the sum of PV and wind currents, while the fuel cell current (Ifc)(I_{fc})(Ifc) is considered as the target output. Based on the EMS logic, the fuel cell remains inactive when renewable generation is sufficient to meet the load demand. When the load demand exceeds the available renewable current, the fuel cell supplies the remaining current according to the battery SOC condition. For low SOC, the fuel cell provides full support; for medium SOC, it provides partial support; and for high SOC, it remains inactive to avoid unnecessary fuel consumption. Four dataset sizes, namely 1000, 3000, 6000, and 10000 samples, are considered to evaluate the learning capability, scalability, and prediction accuracy of the proposed hybrid Ensemble Tree–ANN model.