While modern power sources have improved process stability, real-time monitoring and feedback control remain essential for ensuring consistent weld quality under dynamic conditions. To address this need, a vision-based closed-loop control system was developed for pulsed Metal-Active Gas (MAG) welding. The system dynamically adjusts the welding speed based on real-time visual feedback in the welding process. Otsu thresholding combined with morphological operations was applied to molten pool images for brightness-based feature extraction. These features, representing the dynamic behavior of the molten pool, were incorporated into a feedback loop for real-time control. Without relying on complex model-based prediction or sensor fusion, the proposed method reduces fluctuations in weld bead geometry and lowers the occurrence of defects. The experimental results showed that, under optimized control conditions and after a steady welding state was achieved, the weld bead’s height deviation exhibited an average standard deviation of 0.08 mm, and a process stability rate of 92%. The combination of conventional hardware and straightforward image processing makes the proposed approach practical for industrial implementation.
Pulsed Gas metal arc welding (GMAW) technology is a commonly useful welding technology worldwide, and it’s still a difficult point to keep a safe and stable state in the welding process due to the complex variations. In this study, molten pools serve as a crucial indicator for evaluating welding states. In the pulsed GMAW process, metal transfer predominantly occurs through molten pools. To facilitate effective monitoring and control, a visual-sensor-based control system was developed. By analyzing the physical characteristics of the welding arc under varying welding speeds, real-time images are captured and processed using Otsu’s thresholding method to segment and extract key features. These extracted features are then fed into a closed-loop feedback control system, enabling dynamic adjustment of welding parameters in response to changes in the welding state. Compared with traditional approaches, this study improves welding process stability and reduces the probabilities of weld defects. The results demonstrate that under actively controlled welding conditions, a high-quality weld bead was achieved. The overall average deviation of the weld bead height remained within ± 0.03 mm, and the average variance within ± 0.31 mm², indicating robust control of the welding process. The results maintained 97% of the stability margin. This study provides a foundation for intelligent and adaptive control in automated welding applications.
Real-time management of short-term ozone peaks during arc welding remains challenging because ventilation- and enclosure-defined transport boundaries can create strong position-dependent peak behavior, even under fixed process settings. This study establishes a coordinate-referenced, event-level monitoring and analysis framework to quantify ozone–fume peak coupling under a controlled local exhaust ventilation (LEV) suction boundary during CO2 arc welding. A controlled process–environment testbed with a defined suction condition was implemented, and synchronized ozone and fume signals were acquired at three sampling points referenced to the arc position and the LEV inlet direction. The particulate channel was anchored to a PM4 gravimetric reference, yielding a condition-specific traceable CPM-to-mass conversion factor K1 of 1.76 × 10−2 mg/(m3·CPM) and enabling standardized peak-fume endpoints on a mass-concentration scale. The primary inferential analysis used the curtain-on dataset, comprising 21 sessions and 42 event-level records balanced across three sampling points. Under the same suction boundary, peak coupling was strongly monitoring-coordinate dependent: LEV-aligned locations showed statistically supported ln–ln scaling between peak ozone and peak fume, whereas the opposite-side location did not exhibit statistically supported scaling; a pooled point-parameterized ln–ln model achieved an adjusted R2 of 0.777. As a descriptive control-relevant contrast, adding a curtain enclosure under continuous LEV produced strong event-level ozone peak suppression at LEV-aligned locations, with a maximum reduction of 87.8%, while attenuation at the opposite-side location remained limited. Overall, the results provide a ventilation-boundary-consistent, coordinate-specific basis for monitoring placement and control evaluation, identifying where peak translation is supported and where direct ozone monitoring remains necessary.