Development of Industrial IoT System for Anomaly Detection in Smart Factory

Industrial Internet of Things (IIoT) technology has become a preferred tool for preventing product quality defects, especially in production automation and smart factories. Research in this area has focused on using Internet of Things (IoT) sensors to obtain information regarding various factors, including equipment status. In this study, rather than focusing on equipment status, we focus on identifying factors such as workers’ behavior and working conditions using IoT sensors to improve quality and productivity. To this end, a total of six sensors were used to construct the IIoT sensing device. The R library program was used to analyze the data from the sensors, and a decision tree algorithm was used as an anomalous situation classification model. Out of the 956,404 cases in the data, 256 anomalies were identified. In particular, most anomalies occurred at workstations 4 and 6. Anomalies were detected if the measured variables failed to satisfy the conditions of accelerometer data. In addition, the self-assessment manikin (SAM) model was used to examine the relationship between the working environment and worker well-being. Throat pain index was found to increase with abnormalities in the working environment. Therefore, by comparing the objective data obtained from IIoT and the subjective data from the SAM model, we could confirm that the anomaly detection frequency increases substantially when the operator is unwell.