With the accelerated construction of the power Internet of Things, demand-side resource management has become more informative and intelligent, providing a basis for fine-grained regulation to guide users to save energy and reduce consumption. The traditional cluster load control is mostly oriented to group optimization, and ignores the energy consumption preference of individual users, and it is difficult to meet the differentiated comfort and economic needs of users at the same time. Therefore, this paper proposes a cluster air-conditioning load collaborative control strategy that considering the differentiated energy consumption needs of users. Based on LSTM neural network to simulate user behavior characteristics, this paper introduces the similarity of energy use consumption behavior to measure the degree of compliance with the differentiated needs of users, and then, adopts DQN reinforcement learning to formulate personalized energy consumption strategies, which can reduce user energy costs while meeting the comfort requirements of individual users, and effectively reduce the peak-to-valley difference. Finally, the simulation result verifies the effectiveness and advantages of the proposed strategy.
