Managing Considerable Distributed Resources for Demand Response: A Resource Selection Strategy Based on Contextual Bandit

The widespread adoption of distributed energy resources (DERs) leads to resource redundancy in grid operation and increases computation complexity, which underscores the need for effective management strategies. In this paper, we present a novel approach that decouples selection power dispatch tasks. task determines subset designated participate demand response service, while output selected candidates. A solution strategy based on contextual bandit with DQN structure is then proposed. Concretely, an agent action, solved environment. negative value operational cost used as feedback agent, links two tasks closed-loop manner. Moreover, cope uncertainty problem, distributionally robust optimization (DRO) applied reserve settlement satisfy reliability requirement against uncertainty. Numerical studies demonstrate DQN-based can achieve profit enhancement ranging from 0.35% 46.46% compared policy gradient under different quantities.