Artificial intelligence for predictive maintenance and defect detection

This area addresses Artificial Intelligence (AI) techniques to support smart maintenance in railways. In particular:

  • Novel approaches will be investigated to enable preventive condition-based maintenance by data analytics and machine learning, leveraging on emerging technology like the Industrial Internet of Things (IIoT) for sensing and actuation. In fact, the vast amount of data generated by networked monitoring devices, like drones and Wireless Sensor Networks (WSN), need to be collected and interpreted to generate useful information and knowledge. That allows to replace the traditional approaches based on rigid maintenance schedule and hence improve effectiveness and efficiency in defect detection and problem repair. Bayesian approaches will be investigated to be applied to multi-sensor information fusion for early warning, situation assessment and maintenance decision support.
  • Self-healing approaches will also be investigated based on the paradigms of autonomous and self-adaptive systems.
  • Furthermore, advanced techniques based on Deep Neural Networks (DNN) and Convolutional Neural Networks (CNN) will be investigated to support real-time identification of defects through artificial vision enabled by smart cameras and other smart-sensors.
  • Finally, the Digital Twin paradigm will be applied as a run-time predictive model, based on the multi-formalism integration and cross-checking of diverse AI models.