Arc Image Deep Learning for Real-Time Welding Quality Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current welding quality monitoring technologies, such as those based on frequency analysis of sound and welding current/voltage, are inadequate for accurately assessing welding quality, especially in automated tip-rotating arc welding of aluminum alloys, which are prone to thermal deformation and porosity.
Innovation Solution
A deep-learning-based method and apparatus that generates an arc image-based welding quality inspection model by analyzing images from a camera, identifying welding states, and associating them with quality metrics to improve monitoring accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional frequency analysis methods (sound, current, voltage) are used for welding quality monitoring, then the monitoring system is simple to implement, but the inspection accuracy is insufficient
Solution Approach 1:
The patent replaces conventional electrical and acoustic measurement systems with an optical imaging system. A camera captures arc images during welding, and deep learning algorithms process these images to determine welding quality. This substitution of optical methods for electrical/acoustic methods achieves higher inspection accuracy while maintaining reasonable system complexity
Solution Approach 2:
The patent changes the measurement parameter from electrical/acoustic signals to optical images. By capturing arc images and analyzing visual characteristics through deep learning, the system achieves more accurate welding quality assessment. The parameter transformation from current/voltage frequency analysis to arc image feature analysis resolves the accuracy limitation of conventional methods
2Reliability
If manual welding is used to ensure welding quality, then welding quality can be maintained by skilled welders, but time and cost increase
Solution Approach 1:
The patent implements self-service quality monitoring where the arc image-based deep learning model automatically detects and evaluates welding quality in real-time. The system monitors itself without requiring manual intervention or skilled welder judgment, thereby maintaining high welding quality while improving productivity through automated detection
Solution Approach 2:
The patent establishes a real-time feedback mechanism where arc images are continuously captured during welding, processed by the deep learning model, and used to provide immediate quality assessment. This feedback loop enables automatic quality control without stopping the welding process, maintaining reliability while enhancing productivity
3Manufacturing precision
If tip-rotating arc welding is used for aluminum alloys, then weldability improves and defects reduce, but real-time quality monitoring becomes necessary
Solution Approach 1:
The patent replaces difficult-to-implement electrical/acoustic monitoring with optical image-based monitoring. By capturing arc images and using deep learning for analysis, the system achieves effective real-time quality monitoring of tip-rotating arc welding processes, making the detection and measurement of welding quality more straightforward and reliable
Data Source
AI summary
An apparatus for generating arc image-based welding quality inspection model is disclosed. The apparatus includes: a hall sensor for measuring a welding current flowing in a base metal through an arc welding machine; a voltage meter for measuring a welding voltage through a circuit generated between the arc welding machine and the base metal; a camera for capturing an image of a welding target area on which the arc welding machine performs welding; and a model generator configured to: identify a welding state based on the welding current measured using the hall sensor and the welding voltage measured using the voltage meter; obtain an arc image based on the image captured; associate the obtained arc image with a welding quality identified based on the arc image to generate a dataset; and apply the generated dataset to a deep-learning model to generate an arc image-based welding quality inspection model.