Data-driven Topology Identification Of Low-voltage Distribution Network

The topology information of the distribution network is of great significance to the line loss management.Accurate topological files could help managers analyze the source of line loss and put forward optimization strategies.In order to calculate the management line loss,the accurate connection information between the power supply line,the transformer and the user is indispensable.Abnormal line loss was found in daily management,in addition to "electricity theft",the main cause is often by the inconsistent update of topology document.In practice,this problem relies on the spot to check the topology document,which requires much manpower and material resources and is inefficient.An online search method is urgently needed.The coverage rate of intelligent acquisition equipment is improved,which improves the availability of power measurement data.Mass data could not only provide support for line loss analysis and load prediction,but also for data-driven research on topology of lowvoltage distribution network.According to the topological structure of different power supply links and the measurement data characteristics,this paper analyzes the topological relationship of the low-voltage distribution and carries out research on the error-correction of "line-transformer" relation,"transformer-user" relation and identification of access information of users in low-voltage courts.In order to balance the line load,power dispatching personnel will adjust the topology of distribution network.If document did not do this record accordingly,there will be an error in line loss calculation.In this paper,a data-driven fault correction method for relationship between power supply line and transformer is proposed.Firstly,the voltage data of transformers in different stations of the line to be analyzed are obtained,and the cosine similarity characteristic matrix between transformers is extracted.Then,the matrix is used as input of the abnormal detector.Finally,the abnormal transformers set is generated.The results of the case studies show that the method has high recall and accuracy.Based on the correct "line-transformer" relation,considering "transformer-users" measured data characteristic,this paper puts forward an error correction model,based on power equation relationship between the users and transformer.Considering the power consumption and to the correlation between load and line loss,abnormal user set is given according to the power coefficient,and provides reference for field screening work.Case studies show that the method has a high recall rate and practicability.Further identifying the access information of users can identify and manage the problems such as wrong user files and unbalanced load in low-voltage courts.Considering the characteristics of low-voltage courts,this paper proposes a data-driven identification method for users access information in topology.Input the user voltage data sets in the same court,taking dimension reduction methods to extract user characteristics;Then the unsupervised learning clustering method is used to analyze the user characteristics and identify the phase and access box information of users.Case studies show that the proposed method is feasible and effective.In summary,this paper proposed methods for data driven topology identification in view of the low-voltage courts based on power load data acquisition system,respectively in view of the "line-transformer" relationship correction,"transformeruser" relationship correction and users access information identification,which can provide reference for line loss management,ease the burden on workers and provide convenience for lean management of low-voltage courts.