With the high penetrations of diverse renewable energy resources and energy storage devices, the optimal planning of microgrids based on selected representative operating periods (ROPs) are facing significant challenges. This paper proposes a full data-driven planning method to cope with such challenges. The proposed method takes the full data, which contain all the information on load and meteorological conditions in the planning horizon as input, and applies a two-stage data processing approach to obtain weekly and hourly ROPs, while preserving the chronological order of the raw data. Two time series with different resolutions are established in the planning model in order to satisfy coarse time intervals required for investment decisions and conform to finer time scales required for operating decisions. The proposed method reduces the scale of data, speeds up the solution process, and surmounts the computational burdens in solving the planning model in the designated horizon. Numerical results on an industrial park microgrid in Shanghai demonstrate that the proposed method can provide more accurate ROP and planning results than competitive methods.

中文翻译：

---

可再生能源渗透率高的微电网规划的两阶段全数据处理

随着各种可再生能源和储能设备的高渗透率，基于选定代表性运行期（ROP）的微电网优化规划面临着重大挑战。本文提出了一种完整的数据驱动规划方法来应对此类挑战。该方法以包含规划范围内所有负荷和气象条件信息的完整数据为输入，采用两阶段数据处理的方法获得每周和每小时的ROP，同时保留原始数据的时间顺序。 . 规划模型中建立了两个不同分辨率的时间序列，以满足投资决策所需的粗略时间间隔和运营决策所需的更精细时间尺度。所提出的方法减少了数据规模，加快了求解过程，并克服了在指定范围内求解规划模型的计算负担。上海某工业园区微电网的数值结果表明，与竞争方法相比，所提出的方法可以提供更准确的 ROP 和规划结果。