Integrated Manufacturing Analytics Platform for IoT Enabled Predictive Maintenance
| Reference number | |
| Coordinator | Högskolan i Skövde - Högskolan i Skövde Inst f ingenjörsvetenskap |
| Funding from Vinnova | SEK 6 000 000 |
| Project duration | November 2021 - November 2024 |
| Status | Completed |
| Venture | FFI - Sustainable Production |
| Call | Sustainable production - FFI - June 2021 |
Important results from the project
The project aimed at combining core Industry 4.0 technologies of industrial IoT, digital twins and analytics to realize the full potential of predictive maintenance and pave the way towards prescriptive maintenance. The project goals cover all three typical stages of analytics, namely, descriptive, predictive, and prescriptive. Through six industrial use-cases, the project demonstrated how anomaly detection in predictive maintenance, enabled by IoT sensor data integration, advanced analytics, and machine learning, can improve maintenance operations.
Expected long term effects
The project´s results align with FFI´s Sustainable Production program, which emphasizes digitalization for improved resource efficiency and sustainable production systems. By analyzing time-series and image data from IoT sensors for anomaly detection, the project contributed to reducing unplanned downtimes thereby improving productivity and reducing wastage. The project results are ready for implementation in production and will be passed on to internal development teams at the participating companies. New related research proposals are also being developed.
Approach and implementation
The project addressed six industrial use-cases across three problem domains. Four of them involved anomaly detection and RUL estimation of ball-screws in CNC machines, one dealt with anomaly detection and segmentation in images of sheet metal glue lines, and yet another dealt with hydraulic accumulator fault prediction in a cluster of CNC machines. The methods used in the project include statistical analysis, time-series analysis, signal processing, data integration, classification, image processing, computer vision, deep learning, and multi-criteria decision analysis.