The high penetration of renewable energy sources, reduced system inertia, and dynamic nature of loads present major challenges to frequency stability in modern microgrids. Solar and wind energy generation are intermittent, which leads to significant frequency deviations during normal and disturbed operation. The purpose of this work is to investigate an intelligent load frequency control strategy based on reinforcement learning (RL)-based adaptive PID controllers for renewable-rich microgrids with the aim of increasing their stability, resilience, and sustainability. PID control parameters are continuously tuned with reinforcement learning to optimize frequency regulation under varying operating conditions based on system dynamics. This control strategy is applied to a microgrid that integrates photovoltaics, wind turbines, flywheel energy storage systems, electric vehicles, fuel cells, diesel generators, and battery energy storage systems. As part of sustainable microgrid operation, nonlinearities, uncertainties, and rapid load variations are handled effectively by the controller through learning-based adaptation. The results indicate that reinforcement learning-based PID controller outperforms from various other metaheuristic based PID approaches like grey wolf optimization, salp swarm algorithm and jellyfish search optimization in terms of frequency deviation, settling time, overshoots, undershoots, and integral performance indices. The proposed strategy significantly reduced overshoot by 87.77%, settling time by 64.56%, undershoot by 71.07%, integral absolute error by 91.20%, integral time-weighted absolute error by 69.60%, and integral squared error by 79.03%. Thus, this approach contributes to the development of low-carbon and resilient energy systems through the integration of renewable energy technologies and energy storage technologies.