基于LHS和Cholesky分解的风电出力场景生成研究

Wind Power Output Scenario Generation based on LHS and Cholesky Decomposition

为了有效表征和分析风电出力的不确定性,获得代表性场景,基于风电不同时刻的功率误差,提出了基于拉丁超立方抽样(LHS)和Cholesky分解的场景生成方法。以湖北电网日风电出力为例的研究结果表明,初始抽样结果经过Cholesky分解后,Pearson相关系数方均根降低了39.08%~41.01%,Spearman相关系数的方均根降低了37.68%~40.01%,提高了样本代表性,降低了原始场景之间的相关性,同时验证了方法的合理性和有效性,生成的风电场景可用于后续电网的规划设计与运行调度。

In order to effectively characterize and analyze the uncertainties of wind power output and get representative scenarios,a scenario generation method based on Latin hypercube sampling(LHS)and Cholesky decomposition is proposed considering the wind power errors at different times.Taking the daily wind power output in Hubei Power Grid as an example,studies show that after Cholesky decomposition of the initial sampling results,the sample representativeness is improved and the correlation between the original scenarios is reduced,as the root mean square of the Pearson correlation coefficient is decreased by 39.08%~41.01%,and the root mean square of the Spearman correlation coefficient is decreased by 37.68%~40.01%.The rationality and effectiveness of the proposed method are verified.The generated wind power output scenarios can be used for subsequent grid planning,design,and operation dispatching.