Address resolution is used to locate address information to the urban geospatial system, achieving location-based intelligent vehicle applications and helping to the construction of digital cities. In view of the problems existing in the current field of address resolution that the character embedding model cannot reflect the vocabulary characteristics of polysemy, grammar and semantic features, not suitable for long text, as well as the complex network structure and high operation cost, this paper proposes an address resolution algorithm based on Bert-GRU-CRF. The pre-trained model Bert is used to generate character embedding and BiGRU to build the network, and is improved respectively in terms of accuracy and speed. The dataset in Chinese NLP address element resolution competition (CCKS 2021) is selected to validate this algorithm. The experimental results show that the accuracy, recall, and F1 value of this algorithm are all superior to the conventional methods, as well as the obvious advantages in training time, which has certain theoretical research and practical application value.