Energy Monitoring System for Smart Factory Applications in Textile Industries

Abstract

Many processes in the garment industry, such as the sewing process, still rely on manual labor. Production tracking is crucial not only for determining the current production rate but also for optimizing the process line through line balancing

however, current methods of manually counting finished products are time consuming and prone to error. Other solutions are costly and thus prevent many Small and Medium-sized Enterprises (SMEs) from implementing these technologies. In this thesis, a production tracking system that uses the sewing machines energy consumption patterns for the Convolutional Neural Network (CNN) classifier to track the total number of sewing tasks completed is proposed. This system was tested on two target sewing tasks and was able to detect and count all five tasks. The maximum classification accuracy of the CNN model obtained was 98.6%.