Even in a digital age, welding is an essential method for joining parts, and the typical quality control process is to inspect each part after it’s complete. MxD has a project that uses sensors and artificial intelligence to improve each weld in real time. Not only does this avoid mistakes, it enables the welding process to learn from itself and get better as it goes. This has implications for industries like aerospace, automotive, energy, and shipbuilding – allowing them to avoid waste and scrap and continually improve their operations.
Outcome
The project developed a combined real-time weld quality diagnosis, decision, and control system using multiple sensors and machine learning algorithms. The team used off-the-shelf sensors that feed into algorithms to classify weld quality in real-time, saving the time and money required to use expensive, inaccurate inspection tools off-line. Next steps will involve testing the scanner with different welding equipment to ensure smooth adoption by industry.
Completion Date
Team Members
University of Illinois at Chicago (UIC)
Illinois Tool Works (ITW)
Industrial Measurement Systems
