With the continuous advancement of the digital transformation of the manufacturing industry, traditional industrial equipment faces many challenges in data collection and status monitoring. This paper focuses on the intelligent transformation of sewing machines in the clothing manufacturing industry and proposes a sewing machine working state recognition and analysis method based on edge computing. The current mainstream solution usually relies on uploading the collected sewing machine working data to the cloud for processing. However, this method has strong dependence on the network and has problems such as high bandwidth occupancy, large latency and data loss. To this end, this paper designs and implements an edge computing platform based on the embedded Linux system architecture. The Allwinner V3S chip is selected to realize local real-time processing of sewing machine working data. At the software level, this paper builds an algorithm based on one-dimensional convolutional neural network (1D-CNN) to identify the sudden change points in the working current waveform, thereby judging the completion of the process. The algorithm model is trained using a supervised learning method, by introducing a loss function and applying the gradient descent method to optimize the training results, the model's generalization ability is enhanced while ensuring high recognition accuracy. Results show that the proposed model has a good deployment effect on embedded edge computing devices and has high real-time and stability. This study provides a practical solution for the intelligent upgrading of traditional industrial equipment, which has important practical significance for the digital and intelligent transformation of the manufacturing industry.