Automatic Generation Control Strategy Based on Deep Forest

As the scale of the power system continues to expand and the energy situation changes dramatically, the existing automatic generation control (AGC) strategy needs to be optimized and improved. The current grid AGC mainly adopts closed-loop PI control. By learning an excellent data set that incorporates the characteristics of PI control and DFT control, this paper proposes a real-time AGC strategy based on a deep forest network. The strategy selects the controller with better control performance in each assessment period as the controller for the assessment cycle for power deviation regulation studies. The simulation results show that the strategy can achieve real-time AGC regulation with a lower number of actions and outperform any of the learned strategies.