目前我国正加快构建以新能源为主体的新型电力系统，大量接入的新能源致使弃风弃光现象突出。为提升电力系统新能源的消纳水平，本文提出了一种基于负荷准线和深度强化学习的新能源消纳新方法。首先，本文提出基于线性化潮流计算方法的节点负荷准线形成模型，该模型能够引导可调负荷调整用电时段，从而促进新能源消纳水平的提高。同时，该模型为交流潮流模型，其优势在于，相比于直流潮流模型，它考虑了电压约束等电力系统的相关约束，相比于其他交流潮流模型，该模型将所有非线性化约束进行线性化处理，计算成本更低。其次，本文构建了负荷准线机制的市场框架，并以电动汽车集群为可调负荷，研究了负荷准线激励价格的求解方法。负荷准线机制框架包括独立系统运营商、区域电网售电商以及电动汽车可调负荷聚合商三个主体。负荷准线激励价格的求解涉及三个主体之间的主从博弈，数学解析方法求解难度大，故考虑采用深度强化学习算法求解。深度强化学习算法以各节点边际电价为状态空间，以负荷准线激励价格作为动作空间，以区域电网售电商的成本作为反馈。通过不断地训练智能体，智能体能够找到最大化区域电网售电商利益的负荷准线激励价格。最后，算例分析表明负荷准线机制不仅能够有效促进新能源消纳水平的提高，还能提升包括独立系统运营商、区域电网售电商和电动汽车聚合商的利益。同时算例表明深度强化学习算法能够最大化区域电网售电商的利益。

At present, China is accelerating the construction of a new type of power system with new energy as the main body, and the large amount of new energy connected has led to the prominent phenomenon of wind and light abandonment. To improve the level of new energy consumption in the power system, this paper proposes a new energy consumption method based on load alignment and deep reinforcement learning. Firstly, this article proposes a node load line formation model based on linearized power flow calculation method. This model can guide adjustable loads to adjust the electricity consumption period, thereby promoting the improvement of new energy consumption level. Meanwhile, this model is an AC power flow model, which has the advantage of considering voltage constraints and other related constraints of the power system compared to the DC power flow model. Compared to other AC power flow models, this model linearizes all nonlinear constraints and has lower computational costs. Secondly, this article constructs a market framework for load alignment mechanism, and studies the solution method for load alignment incentive prices using electric vehicle clusters as adjustable loads. The load alignment mechanism framework includes three main entities: independent system operators, regional power grid sellers, and electric vehicle adjustable load aggregators. The solution of the load benchmark incentive price involves a master-slave game between three entities, and the mathematical analysis method is difficult to solve. Therefore, it is considered to use deep reinforcement learning algorithm to solve. The deep reinforcement learning algorithm takes the marginal electricity price of each node as the state space, the load benchmark incentive price as the action space, and the cost of regional power grid sellers as feedback. By continuously training the intelligent agent, the agent can find the load line incentive price that maximizes the benefits of regional power grid sellers. Finally, the example analysis shows that the load alignment mechanism can not only effectively promote the improvement of new energy consumption level, but also enhance the interests of independent system operators, regional power grid sellers, and electric vehicle aggregators. At the same time, examples show that the deep reinforcement learning algorithm can maximize the benefits of regional power grid sellers.