Predictive maintenance using advanced cluster analysis (PACA)
About the project
Predictive Maintenance (PdM) using Artificial Intelligence (AI) and Machine Learning (ML) is the top-ranked use case in terms of business value of industrial digitalization. Not surprising since the annual maintenance cost in Swedish manufacturing industry is over 100 billion SEK and 60% of all maintenance activities are reactive. The PACA project aims to develop PdM algorithms, based on advanced cluster analysis, to increase the precision and make them understandable for decision makers.
Three real-world cases provide data from multiple streams (sensors and computer systems) and multiple machines. The data will be jointly analysed to identify interesting patterns and compare across machines and their historical records. This will build understanding of how different patterns correlate to certain wear-down behaviour, later used to design an algorithm for prediction of future machine states/failures.
Expected effects include: increased productivity, robustness, resource efficiency and competence in Smart Maintenance and advanced data analysis. The cross-disciplinary consortium consists of major manufacturing companies, service and IT providers, and universities with expertise in Smart Maintenance and advanced data science.
The project is a collaboration between Chalmers University of Technology (coordinator), the University of Skövde, AB Volvo, Volvo Car Corporation, SKF, Siemens and IFM Electronic. The project is funded by Swedish Governmental Agency for Innovation Systems (Vinnova) within the Production2030 program and the total grant is 5 million SEK (and the total budget is 11 million) during 2019-03–2022-02 of which the University of Skövde receives 2.25 million.
Se also Vinnova's description of the project.
The figure illustrates two data streams where clustering has been performed in order to identify patterns.