As consumer and industrial energy demands rise, modern power systems are evolving through the integration of microgrids comprising renewable energy sources and electric vehicles (EVs). This decentralization increases the complexity of frequency regulation. This article presents a comprehensive study on a hybrid two-area power system focusing on automatic generation control in a deregulated environment. Area-1 consists of thermal, hydro, and gas units, while Area-2 is a microgrid integrating diesel, wind, solar power, and an aggregated EV model. To ensure robust frequency regulation, a novel cascaded controller structure, \((2\text {DOF-FOPTID}+1)-(1+PI)\) , is proposed. Compared with single-stage and other cascade techniques, this cascade provides extra tuning freedom and better robustness to low-inertia, renewable-driven variability in hybrid microgrids. The controller is compared with conventional and advanced counterparts, including 2DOF-FOPTID+1, FOPTID, (1+PI), and PID. Parameters are tuned using the Modified Walrus Algorithm (MWA), which incorporates opposition-based learning into the original walrus optimization, enhancing exploration, convergence speed, and avoidance of local optima. A comprehensive comparative analysis, conducted with and without the integration of EVs, reveals that the presence of EVs significantly enhances the system’s dynamic performance. Specifically, the integration of EVs leads to a noticeable reduction in peak frequency deviations and a faster settling time, thereby contributing to improved frequency regulation and overall system stability. With EV integration and relative to the next best controller (2DOF-FOPTID+1), the proposed controller reduces overshoot, undershoot, and settling time by 61.82%, 68.79%, and 35.12% in Area-1, and by 32.39%, 33.70%, and 24.95% in Area-2. The effectiveness of the proposed controller is rigorously evaluated under stochastic disturbances arising from variations in load, solar irradiance, and wind speed. Finally, real-time validation is conducted using OPAL-RT in the RT-Lab environment to confirm the practical viability of the proposed controller.
