Under the background of the “double carbon” goal promotion and the energy system reconstruction of industrial parks, facing the intertwined problems of new energy fluctuations, load time-varying and user response uncertainty, this paper proposes an autonomous scheduling method for hybrid demand response of industrial park microgrid under the framework of multi-agent cooperation. Based on the three-layer architecture of data perception, resource coordination and coordination decision, the model integrates dynamic trapezoid-shaped fuzzy modeling of interruptable load, spatio-temporal and behavioral coupling prediction of electric vehicles, day-ahead economic scheduling and intra-day rolling robust correction, and introduces a three-level response mechanism of core-elasticity-emergency. The experimental results show that the average load modeling error of the model is 6.9%, the hot spot matching rate of electric vehicles is 90.0%, the Accuracy and F1 are 93.7% and 92.5%, respectively. Under the four typical scenarios, the average regulation cost is reduced by 24.3% compared with the traditional strategy, the average renewable energy consumption rate reaches 80.3%, and the average power recovery time is shortened to 32 minutes. The research results show that this method is of great significance to improve the economy, flexibility and stability of the park microgrid.
