产品协同设计中异构模型数据融合的有效策略

针对复杂产品设计中，不同设计工具产生的模型数据之间的融合问题，提出了一种工具间端到端的异构模型数据融合策略。利用数据库管理动态特性，通过模型信息共享，实现异构模型数据之间的融合。在OpenMBEE系统集成环境中，通过建模工具CREO二次开发，利用所提策略获取全生命周期设计中的动态模型属性信息，通过3D模型编辑及重用功能测试，验证了所提策略的有效性。利用自动获取可视化模型属性信息的智能算法，设计一种基于Transformer模型与双向长短期记忆(Bi-LSTM)模型相结合的模型属性智能提取算法，利用神经网络的多层感知特性，通过对模型中属性文本信息进行深度学习、特征分析，实现了对异构数据属性信息的自动提取功能。利用CAMEO建模工具设计的需求分析模型构建模型数据集，验证了智能模型信息自动提取功能的有效性。 

Effective strategy of heterogeneous model data fusion in product collaborative design

Aiming at the problem of model data fusion among different design tools in complex product design, the research explores the fusion strategy of end-to-end heterogeneous model data between tools. A multi-layer collaborative strategy of heterogeneous data is proposed, which uses the dynamic characteristics of database management and model attribute sharing to realize the integration of heterogeneous model data. In the system integration environment of OpenMBEE, through the secondary development of the modeling tool CREO, the strategy is employed to obtain the dynamic model attribute information in the whole life cycle design. The effectiveness of the strategy is verified by 3D model editing and reuse function testing. In order to realize the fusion of heterogeneous data, an intelligent algorithm to automatically obtain the attribute information of visual model is explored, based on Transformer model and bi-directional LSTM (Bi-LSTM) model. Utilizing the multi-layer perceptual characteristics of neural network, the algorithm realizes automatic extraction of heterogeneous data attribute information through deep learning and feature analysis of the attribute information in the model. The effectiveness of the intelligent model information extraction is verified by the model data set that is established with the requirement analysis models designed by modeling tool CAMEO.