The maintenance cost accounts for 50-70% of the overall life cycle cost of an asset. These costs are even higher in the case of break-down maintenance and cyclic maintenance activities. Predictive maintenance seeks to prevent asset failure altogether and promise to lower maintenance cost and provide higher reliability and availability. It is defined in terms of the fourth industrial revolution, which is triggered by the sheer amount of data, cheaper sensor technologies, and advancements in artificial intelligence techniques. This talk provides a brief overview of the predictive maintenance approach from the machine learning perspective. Specifically, this talk highlights data collection strategies, problem framing, possible uses cases, and application domains, followed by three case studies of predictive maintenance for railway switches, concrete bridges and bearing motors.
- Room 2 - SMC
- Lecture in English