Predictive Maintenance through the lens of complexity science & statistical model checking

Do you like train delays? Probably not … Maintenance is crucial to keep assets, like trains, up and running. Predictive maintenance is an innovative technique, that provides efficient solutions, by predicting failures (either by AI & data, or physical models or their combination) so that maintenance can prevent these failures.
While predictive maintenance is successful for components, scaling up to system-level is a major challenge. Here is where complexity science comes in, i.e. the study of complex systems (e.g. cities, companies, the human brain) across fields, like biology, urban planning, physics, and economics. In this field, agent-based models play a prominent role: large simulation models to understand how component behavior leads to system level properties. In this talk, I will explain how statistical model checking helps solve two major challenges in agent-based modeling.
Finally, I will show to what extent fault tree models are in fact agent-based models, and what light complexity science can shed on fault trees and other formal models (e.g. SysML).