Research On Power Network Topology Identification Method Based On Big Data

The low-voltage distribution network is the end link connecting users,and its intelligence level directly affects the work efficiency and difficulty of operation and maintenance personnel,as well as the user’s power experience and satisfaction.For a long time,due to the extensive management of low-voltage distribution network and the complex and frequent changes of low-voltage lines,the network topology information in the power distribution area has been missing,imperfect and not updated in time.Due to the lack of information support of topology structure,the management of power grid companies is still relatively backward in terms of troubleshooting,low-voltage management,three-phase unbalance adjustment,and line loss reduction.How to automatically identify and correct the topology information of the power grid and realize the refined management of the distribution network has become a research hotspot in this field.Using the big data information of the metering automation system of the power grid company,this thesis proposes an automatic identification method of the power grid topology in the power distribution area.The specific research contents are as follows:Firstly,in view of the confusion of the relationship between the distribution area and the user,a method for identifying the user-transformer relationship based on Lasso algorithm is proposed.Using the power data of the distribution transformer monitoring terminal and the user’s smart meter,a linear regression model of the power summation is established,and the identification problem of the user-transformer relationship is transformed into the solution of the regression coefficient.According to the size of the regression coefficient,the user-transformer relationship is judged,and the error list of the user-transformer files is generated by screening coefficients close to 0,which provides strong support for the daily line loss abnormal investigation and file management of power supply personnel.The Lasso recognition model has good recognition recall and accuracy when the data sample size is sufficient and the data is complete.Secondly,a phase and meter box identification method based on t-SNE and hierarchical agglomerative clustering is proposed.First,the t-SNE algorithm is used to reduce the dimension of the historical voltage data,reduce the recognition error caused by noise information and redundant information,and improve the recognition efficiency of the model.Then,the hierarchical agglomerative clustering is used to cluster the voltage data after dimensionality reduction,and the user phase is identified.On the basis of identifying the phase,the meter box to which the user belongs is identified according to the characteristics of the voltage data,and finally a four-layer topology structure of "transformer-line-box-user" is constructed.Through example analysis,the efficiency and accuracy of the proposed method are verified.Finally,on the basis of accurately identifying the topological structure of the power distribution area,an online identification method of low-voltage line parameters based on multiple linear regression is proposed.A mathematical model is established according to the voltage drop equation,and the time series data of voltage,current and active power are used to solve the problem by the least square method.Without additional installation of equipment,the impedance parameters of each branch are deduced from the end user side to the transformer side,so as to realize the parameter identification of the whole distribution area.The experimental results show that this method has high accuracy for line parameter identification and has certain promotion value.