适用于配电网三相优化潮流的数据物理融合驱动线性化方法

Hybrid Data-Physical-Driven Linearization Method for Three-phase Optimal Power Flow in Distribution Network

现有配电网三相潮流线性化模型在重载时精度下降较明显,限制了其应用。为此,提出了一种适用于三相优化潮流的改进型数据物理融合驱动线性化方法。首先,基于配电网物理特性得到三相线性潮流模型。然后,采用偏最小二乘回归的数据驱动方法获得线性化误差模型,对物理驱动模型进行修正。与完全数据驱动型方法相比,三相线性潮流模型保留了线路信息,可在优化潮流中用于描述支路约束。与完全物理驱动模型相比,三相线性潮流模型充分利用数据驱动的优化拟合能力来获得线性化误差与节点负荷之间的线性关系。因为在误差修正项中包含更多维的全局信息,所以线性化模型的精度得到提高,保证了重载时所提方法的精度仍足够高。所提方法具有更好的适用性,能够处理各种连接方式的三相变压器和负荷模型以及考虑相间耦合的分布式电源模型。基于IEEE标准算例,将所提方法与其他可用于优化潮流的线性化方法进行对比分析,结果表明所提方法在系统重载时精度依然很高。

The accuracy of the existing three-phase power flow linearization models in distribution networks decreases obviously under heavy load, which limits its application. Therefore, an improved hybrid data-physical-driven linearization method for threephase optimal power flow is proposed. First, the three-phase linear power flow model is obtained based on the physical characteristics of distribution networks. Then, the linearization error model is obtained by using the data-driven method of partial least squares regression to correct the physical-driven model. Compared with the fully data-driven method, the three-phase linear power flow model retains the line information and can be used to describe branch constraints in the optimal power flow. Compared with the fully physical-driven model, the three-phase linear power flow model can obtain the linear relationship between the linearization error and the node load by making full use of the optimization fitting ability of data-driven approach. Because the error correction term contains more multidimensional global information, the accuracy of the linearization model is improved, and the accuracy of the proposed method is still high enough under heavy load. The proposed method has better applicability, which can deal with the three-phase transformers and the load models with various connection modes, and the distributed generator model considering the coupling between phases. Based on the IEEE standard cases, the proposed method is compared with other linearization methods that can be used for optimal power flow. The results show that the proposed method still has high accuracy under heavy load.