中文摘要：

为满足企业对三维工序模型的需求，提出在工艺语义的驱动下，运用证据理论实现二维工序图的特征识别。将识别过程分为搜索潜在特征和对潜在特征进行验证两个阶段。根据当前工艺语义的加工特征的特点，在工序图中获得潜在特征。运用证据理论对潜在特征进行验证，当潜在特征对当前工艺语义特征的信任度最高时，表明该特征即为当前加工特征，对特征相交的情况通过启发式算法来确定其特征基面。以某一零件加工工艺为例，验证了该方法的可行性和有效性。

英文摘要：

To satisfy the enterprise＇s requirement for 3D process model, the feature recognition of 2D process drawing driven by process semantic was realized by using Dempster-Shafer （D-S） theory. This process was divided into two stages： searching potential features and verifying them. According to the machining feature＇s characteristics of process semantic, the potential features were obtained from process drawing, and the potential features were verified by D-S theory. When the confidence degree reached highest in current process semantic feature by potential feature, the feature was determined as the machining feature. For state of features interacting, a heuristic algorithm was used to determine basic surface of the feature. A parrs process technology was taken as an example to verify the effectiveness and validity of this method.