关键词: 数字孪生车间, 生产行为识别, 拓扑图结构, 图卷积网络, 注意力机制

Abstract: Intelligent recognition for production action is the first step during standardizing production process to construct a digital twin workshop rapidly.An approach based on GCN to describe was proposed and the production action in digital twin workshop was recognized.A digital twin feature and an attention Graph Convolution Network (GCN) model by using topological graph structure and GCN model were constructed,and then the digital twin feature was input into the attention GCN model to realize production action recognition.The attention GCN model achieved better accuracy on NJUST-3D datasets,and was useful to build digital twin model for production action.

Key words: digital twin shop-floor, production action recognition, topological graph structure, graph convolution network, attention mechanism