A Game Model Reflecting the Interaction between Supply and Demand of Power System and Its Q Leaming Solution

This paper establishes a hybrid game model of the supply and demand interaction incorporating Stackelberg game on supply-and-demand interaction and evolutionary game on interaction of load aggregators. And a complex network theory is adopted to describe the changing relationship between demand side load aggregators. To solve the complex issue with multiple individuals and numerous constraints, a novel hybrid game reinforcement learning (HGQL) algorithm is proposed. With full utilization of the knowledge matrix information produced by the interactive game relationship, the algorithm can solve the non-convex optimization problem of multi-agent systems with high quality. Finally, simulations are carried out to validate the adaptability and robustness of HGQL for the supply and demand interaction.