A new system uses machine learning and image processing to automatically monitor and improve production line efficiency.
The system aims to improve Overall Equipment Effectiveness (OEE) by accurately collecting production line data. It uses advanced methods like video image processing and deep learning to detect objects and predict production line status. This approach can learn patterns specific to each production line, making it a more effective solution than traditional methods.
Abstract
Overall equipment effectiveness (OEE) is a necessary metric for monitoring and improving production processes in industry [Nakajima, 1988]. In order to make the OEE calculation properly, one needs to digitize the accurate data coming from the production line on the shopfloor, which is a challenge itself. The solution we present provides the accurate collection of production line status and OEE data required for monitoring and decision making. The problems found in existing solutions are overcome with advanced analytical methods such as video image processing and deep learning / machine learning. There are many solutions in the literature using traditional image processing approaches [Dalal, 2005] or machine learning methods [Felzenszwalb, 2010] to solve the object detection problem in the video. In recent years, deep learning methods have also yielded successful results in object detection [Girshick, 2014]. The innovative aspect of the solution we offer is that it is a system that learns patterns that may be different for each production line, and automatically predicts the production line status.
References
- 1.Nakajima, S. (1988), An Introduction to TPM, Productivity Press, Portland, OR.
- 2.N. Dalal and B. Triggs, “Histograms of oriented gradients for human detection,” in CVPR, 2005.
- 3.P. F. Felzenszwalb, R. B. Girshick, D. McAllester, and D. Ramanan, “Object detection with discriminatively trained part-based models,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 32, pp. 1627–1645, 2010.
- 4.R. Girshick, J. Donahue, T. Darrell, and J. Malik, “Rich feature hierarchies for accurate object detection and semantic segmentation,” in CVPR, 2014.