基于CV-KDE的风电并网系统概率潮流计算

PROBABILISTIC LOAD FLOW CALCULATION OF POWER SYSTEM INTEGRATED WIND POWER BASED ON CV-KDE

高比例风电接入电网后加剧了电力系统的不确定性和随机性,用于系统不确定性分析的概率潮流计算就显得尤为重要。而现有的概率潮流计算中所用的风电概率建模存在需要假设参数分布和不能全面考虑各种随机因素影响的缺点,导致潮流计算结果具有较大误差。文中提出采用考虑边界校正的非参数核密度估计建立概率潮流计算中的风电功率概率模型,并引用机器学习中带交叉验证(cross validation,CV)的网格搜索(grid search)法优化核密度估计的带宽参数。所建模型利用网格搜索法将数据在交叉验证中测试训练得到最优解,得到的带宽参数比传统带宽求解法得到的参数更加精确,且使数据的利用更加充分。最后,利用加入风电后改进的IEEE 39节点电力系统进行潮流计算仿真分析,验证了所提核密度估计概率模型和带宽求解方法的准确性和有效性。

Wind power of the high proportion connected to the power grid exacerbates the uncertainty and randomness of the power system,so the probabilistic load flow calculation for system uncertainty analysis is particularly important.However,the wind power probability model used in the existing probabilistic load flow calculation has the disadvantages of hypothesis parameter distribution and the inability to fully consider the influence of various random factors,resulting in large errors in the load flow calculation results.In this paper,the non-parametric kernel density estimation considering boundary correction is used to establish the wind power probability model in probabilistic load flow calculation,and the grid search method with Cross Validation(CV)is used to optimize the bandwidth parameters of kernel density estimation.The model uses the grid search method to test and train the data to obtain the optimal solution,so that the bandwidth parameters are more accurate than the parameters obtained by the traditional bandwidth solving method,and the data utilization is more enough.Finally,the accuracy and validity of the proposed kernel density estimation probability model and bandwidth solution method are verified by load flow simulation analysis of the improved IEEE 39 node power system integrated wind power.