Having an accurate transformer-customer relationship is key for power marketing management and loss management. The traditional identification method of transformer-customer relationship has high cost and is poor in performance, and cannot be applied to the increasingly large-scale low-voltage distribution network. This paper proposes a method to identify the transformer-customer relationship in a low-voltage distribution network. First, the geographic location information of customers is used to realize the initial merging of neighboring customers. Then, based on a Gaussian mixed model (GMM) clustering algorithm, voltage time series data are clustered and divided. The customer segment result is used as the next iteration initial value. Finally, an incidence convolution identification model based on the balance of energy supply and demand is established to identify the connection relationship between the transformers and customers. The method proposed in this paper is studied in an actual low-voltage distribution network. The method has significant advantages in improving the efficiency and accuracy of transformer-customer relationship identification, and has practical application value and offers engineering guidance. This work is supported by the National Natural Science Foundation of China (No. 51907114).