Comparative Study of Health Monitoring Sensors based on Prognostic Performance

Abstract

In the safety critical systems such as industrial plants or aircraft, failure occurs inevitably during the operation, and it is important to prevent this while maintaining high availability. Therefore, a lot of efforts are being directed toward developing advanced prognostics algorithms and sensing techniques as an enabler for predictive maintenance. The key for reliable and accurate prediction not only relies on the prognostics algorithms but also based on the collection of sensor data. However, there is not much in-dept studies toward evaluating the varying sensing techniques based on the prediction performance and inspection scheduling. It would be more reasonable for practitioner to select different cost of sensors based on the sensors’ contribution on reducing the cost on unnecessary inspection or measurement while maintaining its prognosis performance. Thus, the authors try to thoroughly evaluate the cost-effectiveness of the different sensor with respect to sensor resistance to noise. The simulation is conducted to analyze the prediction performance with varying measurement interval and different level of noise during degradation. Then real run-to-fail (RTF) dataset acquired from two different sensors are analyzed to design optimal measurement system for predictive maintenance.

Understanding the Role of Sensor Optimisation in Complex Systems

“…However, there is a lack of in-depth studies on evaluating sensing techniques based on their prediction performance and inspection scheduling. Park et al [ 64 ] addressed the need to evaluate the cost-effectiveness of different sensors by considering their contribution to reducing unnecessary inspection or measurement costs while maintaining prognosis performance. The authors conducted simulations to analyse prediction performance under varying measurement intervals and different levels of noise during degradation.…”