Sensorless-Based Anomaly Detection and Degradation Assessment of Roller Chain Systems

The reliability of mechanical systems in industrial applications profoundly impacts the overall equipment effectiveness (OEE), encompassing quality, efficiency, and operational costs. In the era of Industry 4.0 and intelligent manufacturing, condition monitoring (CM) and predictive maintenance have become pivotal for enhancing the OEE performance. However, the practical implementation of effective maintenance encounters challenges such as (a) data collection and storage, (b) sensor selection and installation, and (c) the development of reliable CM techniques. In the past, much research on CM has focused on bearings and gears, while other types of rotating machinery, such as chain systems, have received comparatively less attention. These systems are widely employed across various applications including conveyor systems, packaging machinery, and power transmission in mobility, agriculture, mining, and production machinery. This work addresses these industrial challenges by introducing an innovative approach utilizing sensorless strategies and motor information-facilitated process data for CM in roller chain systems. Our methodology offers several distinct advantages: (1) We employ a sensorless strategy, utilizing readily available motor-driver-read process data, to eliminate the need for complex sensor selection, installation, and data management. This approach enhances cost-effectiveness and applicability across diverse industrial systems. (2) By leveraging process data from the motor controller, we segment data and integrate data fusion techniques, to provide a comprehensive understanding of the system’s behavior, which enables the improvement of the CM performance. (3) We compare various features for degradation assessment and implement anomaly detection based on k-nearest neighbor (KNN) algorithms, effectively identifying abnormal statuses in roller chains. Through the validation, the developed methodology demonstrates its effectiveness in addressing real-world CM challenges in industrial environments, offering a cost-effective and reliable solution.