A Survey on Data-Driven Techniques of Remaining Useful Life Assessment for Predictive Maintenance of the SystemPages 58-71
Abstract:
To maintain the system in its ideal state, maintenance is essential. In contemporary technological systems, predictive maintenance is a maintenance method that is given more consideration when planning maintenance strategies. Predictive maintenance helps enhance safety, increase asset life, minimize downtime, avoid over-maintenance, lower maintenance costs, and help in the continuous assessment of a system. The fundamental objective of predictive maintenance is to acquire accurate information on when to schedule maintenance based on the system's actual health before it breaks down completely. Remaining useful life (RUL) prediction is in high demand in modern industry because it serves as a foundation for predictive maintenance scheduling. Proper Estimation of RUL is the most important step in predictive maintenance. Under the same or close prediction error, an underestimated RUL is preferable to an overestimated one because an overestimated RUL increases the risk of unexpected shutdowns, which can have disastrous consequences. As a result, it is necessary and significant to properly estimate the RUL, which reduces the possibility of incorrect maintenance scheduling decisions. Several novel methodologies for RUL predictions have been developed based on the distinct scenarios depending on the system. This study highlights a variety of recent research trends toward precise RUL estimation.
Keywords: Remaining useful life, Hybrid models, Machine learning, Probabilistic models, Predictive maintenance
|
PDF 