Weld Quality Prediction Using In-Process Analytics and AI
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional methods for determining weld quality involve destructive testing, which is costly and results in significant financial losses due to the destruction of parts, especially in high-volume production environments.
Innovation Solution
A system utilizing machine learning algorithms, including regression, classification, and artificial intelligence models, to predict weld quality by analyzing weld parameter data from various sensors and actuators during the welding process, reducing the need for destructive testing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If destructive testing is used to determine weld quality, then measurement precision is improved, but loss of substance increases due to destruction of tested parts
Solution Approach 1:
The patent replaces mechanical/physical destructive testing systems with an optical/electronic inspection system using machine learning. The system captures images of welds during the welding process and uses trained machine learning models to predict weld quality parameters such as pull strength, peel strength, and burst pressure without physically destroying the test parts.
Solution Approach 2:
The patent creates a virtual model or digital twin of the weld quality by capturing images during the welding process and using machine learning algorithms to generate predictions about weld strength and quality. This digital copy allows quality assessment without needing to physically test and destroy actual parts.
2Reliability
If destructive testing is performed on a high sample rate, then reliability of quality control is improved, but productivity decreases due to loss of parts
Solution Approach 1:
The patent replaces destructive mechanical testing with non-destructive optical imaging and machine learning analysis, allowing 100% inspection of all welded parts without removing them from production. This maintains high quality control reliability while eliminating the productivity loss associated with destroying test parts.
Solution Approach 2:
The system performs quality assessment continuously during the welding process by capturing images in real-time and immediately analyzing them with machine learning models. This continuous inspection approach maintains production flow without interruption while ensuring every part is quality-checked, unlike batch destructive testing which removes parts from the production line.
3Measurement precision
If destructive testing is used, then measurement precision of weld strength is improved, but loss of time increases due to testing procedures
Solution Approach 1:
The system performs quality assessment during the welding process itself by capturing images in real-time and immediately analyzing them with machine learning models. This preliminary action allows weld quality to be determined before parts move to the next production stage, eliminating separate post-weld testing time while maintaining accurate strength measurement through the machine learning predictions.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
System and methods for using analytics and algorithms to predict weld quality are provided and include a computer having a processor and memory configured to receive weld parameter data generated during a welding process by a welder to join at least two parts with a weld, input the received weld parameter data to a data analytics model to generate at least one predicted weld quality parameter, compare the predicted weld quality parameter with a weld quality parameter threshold, and generate output indicating at least one of: the at least one predicted weld quality parameter and a result of the comparison between the at least one predicted weld quality parameter and the weld quality parameter threshold