System diagnosis and prognosis (or Prognostics and Health Management: PHM) is a key technology to evaluate the current health condition and predict the future degradation behavior of engineered systems under operating conditions.
PHM includes two main approaches: data-driven and model-based. The data-driven approach extracts information from historical data of the performance of a product to capture trends of product health and predict the life using machine learning (or deep learning) techniques. The model-based approach uses knowledge of material properties, product geometry, life cycle loading, and failure models to estimate the life of a product.
The design life of critical industrial systems such as power/petrochemical plants, manufacturing equipment, and transportations can be up to 25 to 40 years. As the safety and reliability of engineered systems becomes important, diagnostic and prognostic methods have received more and more attentions. Key components in industrial machines such as motors, pumps, rotors, valves and power devices are actively studied.
H. Oh, et al., “Failure Mechanisms of Ball Bearings under Lightly Loaded, Non-Accelerated Usage Conditions,” Tribology International, Vol. 81, pp. 291-299, January 2015.
B. Seo, et al., “Fault Detection Method for Solenoid Valves in Urban Railway Braking Systems Using Temperature-Effect-Compensated Electric Signals,” Transactions of Korean Society of Mechanical Engineers, Vol. 40, pp. 835-842, 2016.
H. Oh, et al., “Scalable and Unsupervised Feature Engineering Using Vibration-Imaging and Deep Learning for Rotor System Diagnosis,” IEEE Transactions on Industrial Electronics, Vol. 65, pp. 3539-3549, March 2018.