Development of Anomaly Detection Technology Applicable to Various Equipment Groups in Smart Factory

This study delves into the creation of anomaly detection technology applicable to a range of equipment groups within smart factories. This advanced technology uses high-performance MEMS vibration sensors, edge CMS devices, and PHM platforms to tackle issues such as data imbalance, learning model limitations, complex equipment operating patterns, and real-time processing. It also addresses central server concentration, data cycling problem, various equipment classification, and algorithm operation problems that can arise when implementing systems in the field. Using AI-based vibration detection algorithms, data can be collected at high sampling rates and analyzed in real-time through edge computing, minimizing latency and mitigating server capacity issues compared to cloud-based analytics. The system continually monitors and learns standard performance data from equipment to provide practical solutions that minimize equipment failures and downtimes. The results of this study are impressive, as it has successfully developed anomaly detection framework and PHM systems that are expected to enhance the efficiency and sustainability of smart factories. Furthermore, the study aims to showcase and improve the effectiveness of predictive maintenance in both domestic and international automotive factory production lines. This revolutionary technology will be a key component in smart and software-defined factories and help companies achieve intelligent automation.