用于含风电场的电力系统概率潮流计算的高斯混合模型

Gaussian Mixture Model for Probabilistic Power Flow Calculation of System Integrated Wind Farm

大规模风电并网使电力系统的不确定性显著增加,给电力系统的安全稳定运行带来更大的挑战,用于系统不确定性分析的概率潮流的研究日益重要。该文针对短期风电功率预测误差不对称甚至多峰的概率密度分布特性,提出采用高斯混合模型对预测误差概率密度分布进行拟合。在此基础上,针对各高斯混合模型中子高斯的随机组合结果,采用改进加权最小二乘法计算各组合对应状态变量的概率密度分布。最后,以各组合中子高斯权重系数的乘积为权重,将各组合对应状态变量的概率密度分布加权整合,得到电力系统概率潮流结果。该方法将高斯混合模型与改进加权最小二乘法相结合,很好地拟合了短期风电功率预测误差的概率分布特性,避免了传统加权最小二乘估计中繁琐的迭代寻优过程,大大简化了电力系统概率潮流求解过程。以改进IEEE14节点系统进行算例分析,验证了该方法的准确性和有效性。

Large-scale integration of wind power into grid significantly increased the power system uncertainty, which brings more challenges to the security and stability of power system operation. Therefore the research on probabilistic power flow is very important for system uncertainly analysis. A Gaussian mixture model was proposed to fit the probability density distribution of short-term wind power prediction error. Thi model can deal with the multimodal and asymmetric probability distribution characteristics of short-term wind power prediction error. Then, the improved weighted least square method was used to estimate the probability density distribution of state variables for each random combination result of Gaussian components. Finally, the weight for each probability density distribution of state variables was the product of the weights for Gaussian components in the corresponding combination. The probabilistic power flow of power system was acquired by summing up all the probability density distributions of state variables with their weights. The probabilistic power flow calculation method combining the Gaussian mixture model with the improved weighted least square method, modeling the probability distribution characteristics remarkably and avoiding the complicated iterative optimization process in traditional methods, can greatly simplify the process of power system probabilistic power flow calculation. The simulation results verified theeffectiveness and accuracy of the proposed approach based on the modified IEEE14-bus system.