基于粗糙集和K-均值聚类的概念设计实例检索方法研究

Methodology study on instance retrieval of conceptual design based on roughness set and K-means clustering

针对产品概念设计非常复杂、涉及内容多等特点,提出基于粗糙集的实例推理技术,并将其应用于实例检索。首先在分析了概念设计特征基础上,建立了基于粗糙集的概念设计实例知识表达系统模型,然后通过基于分层聚类的K-means聚类法离散化决策表中的连续属性,利用粗糙集对离散后的决策表进行实例特征属性约简,抽取重要的特征属性作为检索依据,通过权值计算方法得到各特征属性的权值,检索出最为接近的概念设计实例作为设计参考实例。最后以冲压模具实例检索为应用实例,说明了这种实例检索方法的可行性、有效性。

Directed against the characteristics of the conceptual design of product was very complicated and involving many contents etc. , the instance reasoning technique based on roughness set was put forward. Firstly, on the.basis of analyzing the characteristics of conceptual design, the expression system model of instance knowledge of conceptual design was established based on the roughness set. And then by means of the continuous attribute in the diseretized decision table of the k-means clustering method based on delaminated clustering, and utilizing the roughness set to carrying out reduction on the instance characteristic attribute of the straggled decision table. Taking the being colleted important characteristic attributes as the basis of search, and through the calculation method of weighted value, the weighted value of each characteristic attributes were obtained, and take the most adjacent instance of concep- tual design being searched as the referential instance of design. Finally the instance retrieval of a stamping die was being taken as an application instance, the feasibihty and effectiveness of this instance retrieval method were explained.