Abstract: To solve the problem of recognizing both isolated and interacting features of these large thin-wall parts,a Hierarchical Attributed Adjacency Matrix (HAAM) model was put forward based on models' topological geometrical and special relationship information.The CAD model of both part and feature was represented,and a two phrase algorithm based on edge's convexity was designed for feature searching.A strategy based on HAAM was proposed to judge,revise,merge and split a interacting feature,so the interacting feature was recognized by finding out isolated features.The method had been implemented and tested with spacecraft structural parts on Pro/E platform.The experiment results showed the validity and accuracy of the algorithm.

Key words: feature recognition, hierarchical attributed adjacency matrix, complicated interacting features, thin-wall part

摘要： 为解决大型航天薄壁结构件的特征识别问题,提出基于拓扑层矩阵、几何层矩阵、特殊关系层矩阵的三层次属性邻接矩阵模型的特征识别方法,将所有特征分为单一特征、凹连接相交特征和凸连接相交特征进行识别。设计了拓扑层矩阵描述零件CAD模型几何元素间的邻接属性和凹凸性;设计了两阶段的特征搜索算法,从零件拓扑层矩阵中提取所有特征的拓扑层矩阵;以特殊关系层矩阵表达特征间的关联关系,将属于同一凸连接相交特征的特征进行合并和关联;基于凹连接相交特征的几何层矩阵,采用逐层分割的策略,将凹连接相交特征分割为单一特征;基于单一特征的拓扑层矩阵匹配、几何层矩阵匹配,精确地识别单一特征的形状。基于Pro/E平台建立了面向大型航天薄壁结构件的特征识别原型系统。通过实例零件验证了所提方法的准确性和高效性。

关键词: 特征识别, 三层次属性邻接矩阵, 复杂相交特征, 薄壁结构件