P6 - Monitoring of conveyor belt tension for condition monitoring using sensor fusion on embedded devices
- Event
- iCCC2024 - iCampµs Cottbus Conference
2024-05-14 - 2024-05-16
Cottbus - Band
- Poster
- Chapter
- Condition Monitoring
- Author(s)
- M. Jongmanns, S. Devi - Fraunhofer IPMS, Dresden
- Pages
- 133 - 136
- DOI
- 10.5162/iCCC2024/P6
- ISBN
- 978-3-910600-00-3
- Price
- free
Abstract
A conveyor belt was equipped with a sensor to measure vibrations. Based on these measurements, the belt tension should be determined. A low tension leads to slipping of the belt, while high tension strains the bearings of the belt. Both can lead to faster degradation of the mechanical parts of the conveyor belt such as the bearings. Two approaches to evaluate the data using artificial intelligence (AI) models on edge devices were tested. A microcontroller (STM32F4) running random forest classifier could not determine the state with high accuracy, while the trained model suggested during vali-dation, that the accuracy would be >90%. A NVIDIA Jetson Orin Nano running a CNN also could not hold up to the expected accuracies from the theoretical validation of the model. However, using transfer training, the pre-trained model could be adapted to achieve the goal to determine the tension of the conveyor belt.