基于Copula-SVM的短期风电功率场景预测方法

Copula-SVM based short-term wind power scenarios forecast method

风电场输出功率具有随机性、间歇性以及可控性弱等特点,提高风电功率预测精度对含有大规模并网风电的电力系统安全经济运行具有重要意义。基于支持向量机(SVM)建立短期风电功率的均值预测模型,利用Copula函数对多时段风电功率的预测误差进行相依性建模,结合风电功率的预测均值和预测误差相依性结构,形成短期风电功率场景集合,可以直接用于机组组合等决策过程中。基于某实际风电场进行仿真分析,结果表明,考虑预测误差相依结构的场景集合能够包含风电功率实际值曲线,显示了方法的有效性。

Wind farm＇s output is stochastic, intermittent and low-controllable, and improving prediction accuracy of wind power is of great importance for safety and economy operation of power system integrated with large-scale wind power. Support vector machine was used to predict the expectation of wind power, and Copula function was used to model the dependency structure of multi-period wind power forecast errors. By combining the forecasted wind power expectation with dependency structure of forecast errors, the short-term wind power scenario set could be obtained, which could be used to unit commitment directly. Based on simulation analysis of one actual wind farm,it is shown that the scenario set considering the dependency structure between forecast errors is better than the one not considering the dependency structure can contain the line of actual wind power, revealing the effectiveness of the method.