基于相似性修正的风电功率爬坡事件预测方法

Wind Power Ramps Prediction Method Based on Amendment of Similar Events

为了满足爬坡预测所需的长期高精度风电功率预测要求,提出基于相似性修正的风电功率爬坡预测方法。该方法首先根据Granger因果检测法提取数值天气预报(numerical weather prediction,NWP)中对风电功率预测有效的气象变量,保证基于气象变量和统计模型的混合预测模型的可实现性,并以支持向量回归模型作为基本预测模型。其次,结合历史数据分析气象背景相似性与爬坡事件相似性,给出相似爬坡事件的选取机制。考虑到较高精度的风电功率预测可提高爬坡预测的性能,为此,结合相似爬坡的功率变化修正风电功率的预测结果,并由误差指标分析验证修正模型的优越性。最后,对实际算例进行仿真分析,验证基于相似性修正的风电功率预测模型的可行性。

In order to predict wind power ramps with precise long-term wind power, this paper proposed the ramp prediction method based on historical similarity. Firstly, the Granger test method was applied to extract the useful meteorological variables from numerical weather prediction system. It was realizable for the prediction model combining the meteorological variables with statistical model. And the support vector regression model was used as the basic prediction model. On the other hand, analyzing the similarity of historical ramp events and their meteorological backgrounds, a regime for selecting similar ramps was proposed. To improve the precision of wind power ramp prediction, this paper also introduced the history similar ramps for amendment in wind power prediction, and verifies the validity of this method. Finally, taking an actual case for study, the feasibility of the proposed method was proved.