基于AP聚类-跳转持续MC的风电出力时间序列模拟生成方法研究

Research on Simulation Method to Generate Wind Power Output Time Series Based on AP Clustering-Transition and Persistence Markov Chain

模拟生成概率等效条件下的风电出力时间序列对分析未来场景下风电出力不确定性以及提高风电消纳水平具有重要意义。针对传统马尔科夫链存在的问题,文中提出一种基于AP聚类-跳转持续马尔科夫链(Affinity Propagation Clustering-Transition and Persistence Markov Chain,AP-TP MC)的风电出力时间序列模拟生成方法。首先,对历史风电出力数据进行AP聚类,并对每一聚类类别下的风电序列建立相应的状态跳转矩阵;其次,依据马尔科夫链模型,结合风电出力状态持续时间特性,抽样并叠加符合混合高斯分布的波动分量,形成某一聚类类别下的风电出力时间序列;然后,依托类间转移矩阵,模拟生成风电出力时间序列;最后,通过对比传统马尔科夫链法与文中方法生成的风电出力时间序列与历史序列之间的统计指标、概率分布指标及自相关性指标,验证所提方法的有效性和准确性。

The simulation of wind power output time series under the condition of generating probability equivalence is of great significance for analyzing the uncertainty of wind power output in the future scenario and improving the level of wind power consumption.In view of the problems existing in the traditional Markov chain,this paper proposes a generation method of wind power output time series simulation based on AP clustering-transition and Persistence Markov chain.Firstly,the historical wind power output data are clustered by AP clustering,and the corresponding state transition matrix is established for the wind power output sequence under each cluster category.Secondly,according to the Markov chain model,combined with the state duration characteristics of wind power output,At the same time,sampling and overlaying the fluctuation components consistent with the mixed Gaussian distribution,the time series of wind power output under a certain cluster category is formed.Then,based on the transfer matrix between classes,the time series of wind power output is simulated.Finally,the effectiveness and accuracy of the proposed method are verified by comparing the statistical index,probability distribution index and autocorrelation index between the time series generated with the traditional Markov chain method and the proposed method.