In order to identify the assembly action process of workers and prevent the quality problems of assembly products caused by irregular assembly actions of workers,the recognition of assembly actions based on deep learning methods was studied,and an approach to identify assembly actions based on channel attention was proposed,which integrated temporal and spatial information feature network model.The MYO armband sensor was used to collect sEMG signals,and a data set containing a variety of assembly actions was established; a neural network for assembly action recognition was built,and the network model was evaluated.The results show that this method has high accuracy and certain reference value.