Abstract:

To address such issues as low user participation and coarse incentive strategies in demand response of air-conditioning loads in commercial complexes, a dynamic pricing model integrating tripartite Stackelberg game and deep learning is proposed. Firstly, a three-level hierarchical decision framework of power grid-aggregator-user is designed, and a neural network is used to mine the nonlinear function of user load reduction and incentive electricity price. Secondly, an aggregator profit-risk equilibrium model is constructed, and a penalty mechanism and power grid cost function under the elastic constraint of compliance rate are introduced. And then, the optimal subsidy price and load reduction are identified to optimize the subsidy strategy of power grid demand response. Finally, taking a commercial complex as an empirical case, the results show that the proposed model achieved a 94.2% load reduction compliance rate, reduced the user comfort deviation to 0.86, and optimized the peak-shaving cost of the power grid by 57.14% respectively. This study provides a decision-making tool that integrates game-theoretic equilibrium and behavioral interpretability for demand response in energy-intensive buildings, facilitating coordinated resource regulation in new-type power systems.

Key words: district cooling system, tripartite game, demand response, nash equilibrium, electricity market