In modern warfare, the development of artificial intelligence and unmanned systems provides powerful technical support for naval operations. However, with the explosive growth and disorderliness of data, quickly and accurately extracting valuable information from massive datasets has become a significant challenge. This paper proposes a Chinese relation extraction method to address the information extraction challenges in the naval battlefield domain. We designed a feature extraction approach based on the BERT pre-trained model, which effectively captures global features and entity boundary information. To validate our method's effectiveness, we manually constructed a high-quality Chinese Naval Battlefield Relation Extraction Dataset (NRD). Experimental results indicate that our method achieves excellent performance, with precision, recall, and F1 score reaching 92.46%, 92.37%, and 92.41%, respectively. Ablation experiments further demonstrate the feasibility and effectiveness of our proposed model.