Extraction and processing of surface electromyography (sEMG) signals play an important role in medical rehabilitation, artificial limb control and other applications. This paper proposes a new method for sEMG signal processing based on convolutional neural networks (CNN). Firstly, the time-domain feature extraction method is used to extract the features of sEMG signals. Then, a CNN model is constructed for identifying and processing sEMG signals. The parameters of the model are calibrated based on the sEMG dataset to find the most suitable model parameters for the objective of this paper, achieving optimization of the model. The constructed CNN network is applied to NinanPro dataset for simulation and experiment, and the results show that the optimized CNN model proposed in this paper has a high accuracy rate for sEMG signal recognition.