Objective Addressing the issues that knowledge dependency in high-speed train bogie configuration design is difficult to express and that knowledge retrieval and reuse are insufficient, a high-speed train bogie configuration design method based on a knowledge graph Q&A （question answering） model is proposed.

Method Based on an ontology model, a knowledge graph for high-speed train bogie configuration design is constructed. An entity recognition method for designer queries using the Transformer-based RoBERTa （robustly optimized BERT pretraining approach）–BiLSTM （bidirectional long short-term memory）–CRF （conditional random field） model is proposed. A question intent classification method based on template matching is adopted to establish knowledge graph query statements, thereby constructing a knowledge graph Q&A model that integrates 'question entity recognition + question intent classification.' The knowledge graph Q&A model is embedded into the bogie configuration design process, and a prototype knowledge Q&A system for high-speed train bogie configuration design is developed to assist designers in the configuration design process.

Result & Conclusion  The configuration design of partial bogie modules is used as an example for verification. The precision, recall, and F₁-score of the RoBERTa-BiLSTM-CRF model are 88.51%, 89.23%, and 88.87%, respectively, all of which outperform the comparative models. The developed prototype system supports retrieval based on configuration parameters and can visually display physical module instances that meet the requirements. The proposed method enables the effective management and reuse of bogie configuration design knowledge, providing knowledge support for module selection and design decision-making.