In this project, a novel autonomous control framework Grid Mind is developed for secure operation of power grids based on cutting-edge artificial intelligence technologies. The proposed platform provides a data-driven, model-free and closedloop control agent trained using deep reinforcement learning algorithms by interacting with massive simulations and/or real environment of a power grid. The proposed agent learns from scratch to master the power grid control problem purely from data. It can make autonomous control strategies to support grid operators in making effective and timely control actions, according to the current system conditions detected by real-time measurements from SCADA or phaser measurement units. Various state-of-the-art DRL algorithms, e.g., deep Q-network (DQN) and deep deterministic policy gradient (DDPG) etc., are utilized and analyzed.