基于数据驱动的微电网双层鲁棒优化调度

Double Layer Robust Optimal Dispatching of Micro-grid Based on Data-driven

提出了一种基于数据驱动的微电网两阶段自适应鲁棒优化调度方法。首先,构建了基于数据驱动的微电网市场调度优化框架,利用K-means聚类方法对微电网大量历史数据进行聚类预处理,选取典型场景代表大量复杂场景来获得准确地风电出力概率分布情况。然后,在阶段一建立微电网日前预调度模型;基于预处理的聚类数据,建立数据驱动的风力发电的不确定性集合,在阶段二建立微电网实时调控模型。通过数据驱动构造的风力发电不确定性集合,排除了部分极端场景,降低了模型的保守度。接着,用列约束生成算法(C&CG)将构建的两阶段自适应鲁棒优化模型分解为主问题和子问题进行交互迭代求解。最后,仿真结果验证了所提方法的有效性,降低了微电网设备运行成本,提高了新能源利用率。

A two-stage adaptive robust optimization(ARO)dispatching method based on data-driven micro grid was proposed.First of all,a data-driven market scheduling optimization framework for micro-grid was built.K-means clustering method was used to conduct clustering pretreatment for a large amount of historical data of micro grid,and typical scenes were selected to represent a large number of complex scenes to obtain accurate probability distribution of wind power output.Then,the day-ahead scheduling model of the micro grid was established in stage 1.Based on the pre-processed clustering data,the uncertainty set of data-driven wind power generation was established,and the real-time regulation model of micro-grid was established in stage 2.Through the uncertainty set of wind power generation constructed by data driving,some extreme scenarios were eliminated and the conservatism of the model was reduced.Then,the two-stage adaptive robust optimization model was decomposed into an interactive iterative solution for the main problem and the subproblem by using the column constraint generation algorithm(C&CG).Finally,simulation results verify the effectiveness of the proposed method,it can reduce the operating cost of micro-grid equipment,and improve the efficiency of new energy.