Parameter
Mehr zum Buch
In recent decades, laser beam welding (LBW) has surpassed older welding techniques in various manufacturing sectors, including automobile production and precision mechanics. This advancement has been accompanied by the development of on-line monitoring systems utilizing photodiodes and CCD/CMOS cameras. However, commercially available sophisticated equipment for real-time LBW process control remains elusive, as existing systems lack the necessary spatial and temporal resolution. To address the high dynamics of LBW and ensure robust feedback systems against physical fluctuations, control rates in the multi-kilohertz range are essential. This work focuses on implementing high-speed Cellular Neural Network (CNN)-based algorithms for feature extraction from coaxial process images, specifically analyzing the full penetration hole (FPH) for real-time laser power control and monitoring spatters for welding quality assessment. CNNs consist of processing units called cells that interact with neighboring units, facilitating continuous-time dynamics. The algorithms have been deployed on a CNN-based chip named QEye, part of Anafocus’ Eye-RIS vision system, which enables simultaneous image sensing and processing. This capability, combined with the real-time signal processing power of CNNs, makes Eye-RIS VS an effective development platform. The system has been integrated into a closed-loop control framework, achieving real-time LBW process
Buchkauf
Visual control of laser welding processes by cellular neural networks, Leonardo Nicolosi
- Sprache
- Erscheinungsdatum
- 2012
Lieferung
Zahlungsmethoden
Keiner hat bisher bewertet.