Intelligent Frequency Control in Multi-Microgrids Using Adaptive Neuro-Fuzzy Inference System (ANFIS) With Renewable Energy Sources and Electric Vehicles

Abstract

Frequency stability is a major issue in renewable-rich multi-source microgrids and aggregated microgrid clusters, especially when intermittent RESs, energy storage systems, and EV/V2G units participate in active-power balancing. This study proposes an ANFIS-based intelligent controller for frequency regulation in an equivalent single-area multi-source microgrid model. The proposed framework combines the interpretability of fuzzy logic with the adaptive learning capability of neural networks, thereby ensuring responsiveness to load variations, environmental perturbations, and dynamic uncertainties. The microgrid model used for the study includes wind power, PV systems, diesel engines, combined heat and power (CHP) plants, battery banks, flywheels, and EVs with vehicle-to-grid (V2G) functionality. The efficacy of the proposed controller is evaluated through five main operating case studies, including multi-stage load disturbances, natural-gas pressure reduction affecting CHP output, power-system parameter variations, renewable-generation fluctuations, and output-power coordination among EV, diesel generation, and flywheel units. In addition, supplementary robustness assessments are conducted for heterogeneous multi-type EV charging and balanced/unbalanced fault-induced active-power disturbances. Comparative simulations against a conventional fuzzy controller show that the proposed ANFIS controller reduces the RMS frequency deviation by approximately 41.5%–49.1% and the maximum frequency deviation by 9.7%–43.7% across the investigated operating scenarios. Under the external 0.3 p.u. benchmark disturbance, the proposed controller also achieves about a 58% reduction in RMS frequency deviation compared with the ICA-Fuzzy and FO-Fuzzy-PID controllers, while maintaining competitive peak-deviation performance.