Alexander Karlsson
Associate Professor of Informatics
The project investigates how industrial companies can predict maintenance needs even when there is limited data on new machines and components.
"Research can be used to avoid costly stop in production."
Alexander Karlsson, Associate Professor of Informatics
Predictive maintenance aims at predicting different types of maintenance needs well advance in time so that one can plan and also avoid costly stop in production.
The predictions are often based on data analysis techniques from the research area of artificial intelligence and in many cases those techniques might require quite substantial and also rather specific type of data.
Systems that evolve
But what if a new type of machines / components has been bought in and there is not much data yet. Should one then wait with building predictive models until more data exists? And how should one really work with models and data from a system perspective in that type of situation?
These are some of the questions addressed in the project. The goal is to develop a software prototype system that evolves over time.