With the massive penetration of renewable energy, traditional reinforcement learning algorithms suffer from slow convergence and area control error (ACE) in interconnected power systems. This paper proposes data-driven ...