Predictive Maintenance: Let Data Maintain the Model

Podcast (in Dutch)
During the workshop podcast series about data-driven maintenance (7 episodes) were recorded together with De Dataloog, NL's largest podcast channel about AI.
It is available through all major podcast channels.

- Episode 1: https://open.spotify.com/episode/2JSS9w3WbEuj6v0PlUFta0?si=v2EnqDkhSRKxhC5aez-Xmg

- Episode 2: https://open.spotify.com/episode/4vX1DC5YyQ9jc7hNDiLNco?si=sV8n7-KFRpi50uTvM24eOg

More episodes will follow.

----------------------------------------------------------------------------------------------------------------------------------

Predictive maintenance (PdM) is an advanced maintenance concept that employs data-driven analytics to optimise the availability of assets. PdM works by exploiting the large amount of data collected for industrial systems and advanced AI methods. The gathered data is analysed to create accurate models of the real-life system which can be used to predict future behaviour. These predictive models then allow pinpointing the optimal maintenance policies. “Optimal” here means only applying inspections and repairs if truly necessary, but still guaranteeing safe and reliable system operations. PdM therefore performs targeted maintenance and allows to reduce costs while not compromising safety.

The aim of the meeting is to bring academics and practitioners with different backgrounds e.g., AI, formal methods, asset management, monitoring, and sensor systems together to discuss the main challenges of predictive maintenance (PdM). In particular, we discuss how to connect the individual PdM steps in an overarching PdM pipeline. We will discuss challenges, emerging topics, best practices, and we aim for concrete initiatives for joint research.