Identification and correction of meter-transformer associations in residential LV networks

The growing adoption of residential distributed energy resources needs advanced analyses to assess their impacts, which often depend on detailed low voltage (LV) electrical network models. However, distribution companies face challenges in ensuring high-quality databases for producing such models. One critical shortcoming is the inaccurate association of meters (customers) with their corresponding transformers (i.e., meter-transformer associations). Imprecise meter-transformer associations result in flawed network models, hindering effective planning and operation. This paper presents a practical two-stage methodology to identify and correct false meter-transformer associations in databases by exploiting smart meter voltages. Principal Component Analysis and Agglomerative Nesting techniques are used to identify false associations, which are then corrected by reassigning to their actual transformer using a correlation-based method. The effectiveness of the methodology is demonstrated through a case study with three scenarios. Each scenario purposely introduces false meter-transformer associations, mimicking real-world data quality issues. The case study uses a realistic network from Victoria, Australia, with synthetic smart meter data of 30-minute resolution over two weeks. The results show the accurate identification and correction of false associations within large LV networks (over 100 customers). This highlights the potential of the proposed methodology to help distribution companies improve the accuracy of meter-to-transformer information.