Traditional transformer topology identification methods have poor accuracy and efficiency and are easy to be interfered. In order to solve the problem, this paper proposes a transformer topology identification based on LoRa and GPU acceleration, which aims to obtain and analyze the data from large quantities of smart meters and effectively identify the corresponding relationship between transformers and meters, via LoRa communication technology, high performance computing technology and large data methods. This paper adopts high density data acquisition method based on LoRa and protocol compression techniques-‘Fast Acquisition and Slow Delivery’, which effectively strengthens the fast acquisition of data. Meanwhile, data analysis is implemented by GPU parallel accelerated grey correlation analysis method, which effectively improves the efficiency of the method. Numerical experiments show that the method proposed in this paper has high accuracy and computational efficiency, which has the value and potential of engineering applications.