基于智能电表量测的配电网线路异常参数辨识及定位方法

Abnormal Parameters Identification and Location Method of Distribution Network Line Based on Smart Meter Measurement

为提高配网线路参数质量,提出一种基于智能电表量测的配网线路异常参数辨识及定位方法。该方法把传统辨识算法的非线性辨识方程求解问题转化成参数最优分布的推断问题,进而在参数辨识的基础上,利用概率统计法进行异常参数定位。首先,给定线路参数初始分布,利用马尔科夫链蒙特卡洛方法生成参数样本,并通过树状结构估计方法和损失函数进行参数分布更新,并以损失函数结果收敛时的参数分布期望作为线路参数辨识值;其次,计算线路参数相对偏移距离,通过概率统计方法判断辨识数据是坏数据或异常参数,并将坏数据直接剔除;最后,计算分析反映线路参数错误的异常因子,进行线路异常参数定位。通过实际29节点的10 kV馈线展示了参数辨识的流程,并通过实际97节点的10 kV馈线进行异常参数定位,证明了所提方法的可行性与有效性。

To improve the quality of power distribution network parameters,an abnormal parameter identification and localization method for distribution networks based on smart meter measurements was proposed.The method transformed the nonlinear identification equation solving problem in traditional identification algorithms into the inference problem of the optimal distribution of parameters.On the basis of parameter identification,probability statistics method was used to locate abnormal parameters.Firstly,given the initial distribution of line parameters,Markov Chain Monte Carlo method was used to generate parameter samples.The parameter distribution was updated through tree estimation method and loss function.The expectation of the parameter distribution when the loss function converges was taken as the identified value of the line parameters.Secondly,the relative deviation distances of line parameters were calculated,and probability statistics method was used to judge whether the identified data are bad data or abnormal parameters.The bad data were directly eliminated.Finally,the abnormal factors causing the incorrect feedback of line parameters were analyzed to locate the abnormal parameters of the line.The identification process of parameters was demonstrated through an actual 29-node 10 kV feeder.The abnormal parameter location was carried out through an actual 97-node 10 kV feeder,proving the feasibility and effectiveness of the proposed method.