基于微型PMU的配电线路抗差参数辨识

Robust parameter identification of distribution line based on micro PMU

微型相量测量单元装置为智能配电网提高可观性和可控性提供了基础。研究了基于微型相量测量单元的配电网线路参数抗差辨识方法:针对配电网三相运行不对称的情况,构建了基于相分量描述的配电线路三相π型等值模型,结合微型相量测量单元提供的多组线路两端电压、电流相量数据,提出了基于相分量模型的、适应配电网不对称运行的最小二乘线路参数辨识方法;并引入2种抗差方法,一种为基于Huber估计的抗差最小二乘法,另一种为基于中位数估计的抗差最小二乘法。在中性点经消弧线圈或小电阻接地系统中,所提方法仅需要线路两端的电压和电流相量即可在辨识线路相参数的同时获得线路正序和零序参数;在中性点不接地系统中,由于没有零序电流分量,当系统正常运行时所提方法可辨识线路正序参数。算例仿真验证了所提方法的有效性,并比较了2种抗差方法在辨识精度、计算耗时方面的优劣,结果表明,基于中位数估计的抗差最小二乘法的抗差性能更好,计算时间较短。

Micro PMU( Phasor Measurement Unit) provides the foundation for improving the observability and controllability of intelligent distribution network. A robust identification method for the distribution line parameters based on micro PMU is studied. Aiming at the asymmetric operation of three-phase distribution network,the three-phase π-type equivalent model of distribution lines based on phase component description is established,and then the line parameter identification method based on the LS( Least Squares) algorithm is proposed considering multiple sets of voltage and current phasor data at both ends of lines provided by micro PMUs,which is based on the phase component model and is suitable for asymmetrical operation of distribution network. And two robust estimation methods are introduced,one is HRLS( Huber estimation based Robust Least Squares) algorithm and the other is MRLS( Median estimation based Robust Least Squares) algorithm. In the system whose neutral point is grounded by arc suppression coil or low resistance,the proposed method only requires the voltage and current phasor at both ends of lines to identify the phase parameters and obtain the positive and zero sequence parameters of lines. In the neutral ungrounded system,the proposed methods can identify the positive sequence parameters of lines when the system is in normal operation since there is no zero sequence current component. The effectiveness of the proposed method is verified and the advantages and disadvantages of two robust methods in identification precision and computing time are compared by example simulation. Simulative results show that MRLS algorithm has better robust performance and less computing time.