基于聚类抽样的随机潮流计算

Stochastic Load Flow Calculation Method Based on Clustering and Sampling

随机潮流针对电网不确定因素建模,快速准确的计算结果对于系统运行控制非常重要。该文提出了基于聚类抽样的随机潮流计算方法。首先,基于历史数据,采用蒙特卡洛算法产生大量随机变量样本;其次,采用误差二次方和与平均轮廓系数确定样本聚类簇数,并采用Kmeans算法进行样本聚类;然后,依据聚类中心和簇内样本的平均概率密度,将簇内样本均值带入潮流方程中进行确定性潮流计算;最后,对潮流计算结果和对应的平均概率密度信息进行数据统计,得到状态变量的概率密度函数。以修改后IEEE 39系统和某实际区域电网为算例,进行计算精度和计算效率分析。结果表明,所提方法兼顾了计算速度和计算精度,在大系统中更具优势,可为电网调度计划的制定和运行分析提供决策依据。

Stochastic load flow is designed for power systems with uncertainties,whose fast and accurate calculation results are very important for grid operational control.In this paper,a stochastic load flow calculation method was proposed on basis of clustering and sampling.Firstly,according to history data,Monte Carlo simulation method was used to generate a large number of random variable samples.Secondly,the optimal cluster number was determined for samples by the average silhouette coefficient and sum of squared error,and the samples were clustered by using K-means.Thirdly,according to the clustering center and the average probability density of the sample in the cluster,load flow was calculated with the mean values of each cluster.Finally,the power flow calculation results and the corresponding average probability density were statistically analyzed,and the probability density function of the state variables was obtained.The modified IEEE39 system and a real regional power grid were taken as examples to analyze the calculation accuracy and calculation efficiency.The results show that the method proposed in this paper can balance calculation accuracy and calculation speed,and there are more advantageous in large systems.It provides decision basis for grid dispatching plan and operation analysis.