计及需求响应柔性调节的分布鲁棒DG优化配置

Distributionally Robust Optimal DG Allocation Model Considering Flexible Adjustment of Demand Response

弃电现象制约着风电、光伏的发展,采用需求响应措施提高配网对清洁能源的消纳是一种有效的措施,但传统处理分布式电源(distributed generation,DG)出力或需求响应不确定性的方法具有很大的局限性,因此,该文建立计及需求响应柔性调节的配电网两阶段分布鲁棒DG优化配置线性模型。首先,该模型以配电公司运营商年利润最大化为优化目标,考虑各种投资约束和运行约束,并将清洁能源出力和负荷的差值作为基于实时电价的需求响应的定价依据,以提高配电网对清洁能源的消纳率;其次,通过基于多面体的线性化技巧和McCormick方法,将原始混合整数非线性模型线性化,建立混合整数线性规划模型;然后,充分利用需求响应和DG出力的历史数据,构建数据驱动的两阶段分布鲁棒DG优化配置模型,其中第一阶段确定DG的投资方案,第二阶段模拟投入DG后的系统运行,并同时考虑不确定性概率分布置信集合的1-范数和∞-范数约束;最后,采用列与约束生成(columnandconstraintgeneration,CCG)算法对分布鲁棒模型进行求解,并基于IEEE33节点系统验证所提模型的有效性。

The development of wind power and photovoltaic is mainly restricted by wind and solar curtailment.Demand response(DR) is an effective measure to improve the consumption of clean energy for distribution system.Traditionally, however, it has great limitations on the method of dealing with the uncertainty associated with distributed generation(DG) output or DR. A two-stage distributionally robust DG optimal allocation linear model considering DR flexible adjustment was formulated. The annual profit of distribution system operator(DSO) considering various investment and operational constraints was maximized. In order to improve the consumption of clean energy for distribution system, the author regarded the difference between clean energy output and load as the pricing basis of DR based on real-time price(RTP). Then, through the polyhedral-based linearization and McCormick’s method, the original mixed integer nonlinear programming(MINLP) model was converted to a mixed integer linear programming(MILP) model.Moreover, taking advantage of the historical data of DR and DG output, a data-driven two-stage distributionally robust DG optimal allocation linear model was presented. The first stage determined the investment plan of DG, and in the second stage,the system operation stage after investing DG, the 1-norm and∞-norm constraints of the uncertainty probability distribution confidence set were incorporated. Finally, the proposed model was simulated on the IEEE33-node system and solved by column and constraint generation(CCG) algorithm. Numerical results demonstrate the effectiveness of the proposed model.