风电功率爬坡事件预测时间窗选取建模

Modeling on Selection of the Time Window for Ramp Events Prediction

为了解决多长时间段的风电功率预测数据能够有效地完成爬坡事件预测,提出爬坡预测窗口的选取问题。假设以一个预测窗内能完整预测一个爬坡事件为目标,根据预测时间窗内冗余信息最小建立数学规划模型。然后,结合历史爬坡事件持续时间的统计分析和风电功率数据本身的可预测性研究,分析模型最优解的存在性并给出模型的参数约束范围和实现算法。其次,以美国BPA区域的风电功率数据为实例,根据具体的模型求解过程,仿真得到模型的最优解,并以该最优解作为预测时间窗的选取。最后,根据预测评估指标验证该时间窗具有较好的预测性能,从而说明该预测窗口对爬坡事件分析的有效性,和该预测窗口选取模型用于爬坡事件预测的可行性。

In order to solve the problem that how many data should be required to predict future wind power ramps, this paper put forward a method for selecting the optimal time window. Assuming that a ramp event could be completely forecast in a prediction window, the paper established a mathematical model based on the optimal redundant time in a window. Then, combining the statistical analysis of historical ramp events duration and the study of the predictability of wind power data, the upper and lower bounds of the window size for forecasting ramps were given out. Taking the wind power data from BPA area of United States as an example, the optimal solution was obtained by simulation as the given process and regarded as the time window for ramp prediction. Finally, several evaluation indicators were used to verify that the selected time window is valid for ramps analysis and that the model for the selection of prediction window is useful to forecast ramp events.