Machine learning applications in investigation of the processes of transformer manufacturing plant

To ensure the optimal operation of industrial or service businesses, effective time management is crucial. Traditionally, time calculation has been performed manually using a stopwatch, leading to a significant waste of time. In this study, we automated the calculation of the active working time for assembly line operators in the Switching Products Breaker Production Department of a transformer manufacturing facility. We identified factors influencing the operators' work and provided suggestions to enhance the efficiency of the production line. The YOLO (You Only Look Once) algorithm, a popular object detection algorithm known for its speed and accuracy, was utilized. This algorithm, based on Convolutional Neural Networks (CNN), processed images captured by a video camera on the production line, allowing for the tracing of operators. The collected data underwent analysis through ANOVA tests and Regression Analysis to examine the factors influencing productivity. The results revealed that working hours significantly impact employee productivity. Conclusively, this study recommends efficiency improvements based on automatic time calculation using object detection.