Predictive maintenance continuously analyzes the condition of equipment during normal operations to reduce the likelihood of future failure.
Another possibility is predictive maintenance: here the goal is to improve machine reliability and better service planning either in supervised or unsupervised way. The focus of the analysis can be deviation detection -- either by specifying the actual failure or notifying about some expected anomaly -- or estimation of remaining useful life.
Examples for maintenance techniques
The challenge at this point, again, is the acquisition of properly labelled historical data on one hand, and the complex, interdependent nature of the systems. Here may different models be needed for different subsystems, and different kind of data -- either from fault detection systems or service data bases -- are required.
Apart from these immediate goals further analyses can be derived from all these data could give novel insights for stakeholders in the form of prediction to, among others, energy consumption prediction, real-time route optimization.
Added values (Why AI/ML/DL): prediction of sudden vehicle breakdown that make prevention possible.