Abstract: Aiming at the problem that the searching,changing and cost estimating of similar cases could not be realized when the case-based design method was applied to complex product,a method to construct the mapping relationship between function attribution and global feature attribution was proposed by building the global feature attribution.The global feature attribution was defined by geometry attribution,material attribution and manufacturing information.Based on multi-level artificial neural network,the mapping model between function attribution and global feature attribution was built.On this basis,the models of intelligent design and cost estimation were proposed,which could realize the physical structure similarity evaluation,autogeneration of case adaptation and new cost estimation.The methods of global feature attribution description,mapping model construction and distance-based similar case searching were introduced.The feasibility of the proposed model was verified with cases,and the technology support was provided for intelligent design of complex products.

Key words: intelligent design, cost assessment, global feature attribution, complex product

摘要： 针对基于实例设计方法应用于复杂产品时难以实现相似实例搜索、修改和产品成本估算的问题,提出一种通过建立产品的全局特征属性建立产品功能属性与全局特征属性映射关系的方法。全局特征属性包括产品结构的几何属性、材料属性和制造属性。基于多级神经网络建立功能属性与全局特征属性之间的映射模型,并在此基础上建立了适合复杂产品的智能设计和成本估算模型,实现了复杂物理结构相似性评价、适应性修改方案自动生成和新产品的成本估算。论述了全局特征属性的描述方法、映射模型的构建方法和基于距离的相似实例搜索方法。结合实例验证了模型的可行性,为复杂产品智能设计提供了技术支持。

关键词: 智能设计, 成本估算, 全局特征属性, 复杂产品