AIVIONICS - AI-Driven Weld Quality Assessment in Metal Manufacturing
Oxiplant is a Spanish metal services company based in Asturias, with more than 20,000 square meters of production facilities. The company specializes in metal cutting and transformation processes, including laser cutting, plasma cutting, and fabrication services for customers in the automotive, steel, renewable energy, and industrial sectors.
As part of the AVIONICS experiment within AIRISE, Oxiplant sought to improve the quality assessment of its MIG welding operations. By leveraging artificial intelligence and computer vision technologies, the company aimed to automate weld inspection processes, improve defect detection accuracy, and support the digital transformation of its manufacturing operations.
Smart Manufacturing Quality Assurance
AI-driven Automated Weld Inspection
As Oxiplant expanded its welding capabilities, the company needed a reliable and automated method for assessing weld quality. Traditional inspection techniques did not fully meet the requirements of its production environment, creating challenges in maintaining consistent quality standards.
The company aimed to introduce an AI-based quality assessment system capable of automatically detecting welding defects from high-definition images. However, implementing such a solution required overcoming technical challenges related to model selection, real-time processing requirements, and the limited availability of high-quality training data.
In particular, the scarcity of labeled defect images made it difficult to develop a robust AI model capable of operating effectively in industrial conditions.
Within the AIRISE AVIONICS experiment, Oxiplant implemented an AI-powered weld inspection solution based on neural network technology and computer vision techniques.
The system analyzes high-definition images of welds to automatically identify porosity defects, one of the most common welding quality issues encountered in the company's production processes. To overcome the limited availability of training data, the project team applied data augmentation techniques to generate additional synthetic images and improve the robustness of the AI model.
AIRISE experts supported Oxiplant throughout the development process, providing guidance on model selection, dataset preparation, and training methodologies. Through the Processing, Monitoring and Control AI service, the company was able to validate and optimize the solution for industrial use.
The resulting system enables automated quality assessment and provides a foundation for future expansion to additional welding defects and production environments.
- High-Accuracy Automated Defect Detection:
- The AI-powered inspection system successfully identified porosity defects in welded components using image-based analysis.
- Validation tests demonstrated an accuracy of approximately 98% in porosity defect detection.
- The solution achieved a processing throughput of up to 12 images per second, demonstrating its suitability for industrial applications.
- Improved Quality Control Processes:
- Automated inspection provides a more consistent and objective approach to weld quality assessment.
- The system reduces reliance on manual inspection activities and supports faster quality verification processes.
- The successful validation of the solution demonstrated the feasibility of integrating AI into welding operations.
- Accelerated Digital Transformation:
- The project introduced new AI capabilities into Oxiplant manufacturing processes.
- Employees gained first-hand experience with artificial intelligence technologies and their practical industrial applications.
- The experiment helped establish a foundation for broader adoption of AI across the company's operations.
Following the success of the AVIONICS experiment, Oxiplant plans to deploy the AI-based quality assessment solution across its production lines as an internal service.
Future developments include integrating the technology into both manual welding operations and welding processes performed by collaborative robots. The company is also exploring additional AI-driven applications in areas such as automated quotation generation and production cost estimation.
The positive results achieved through AIRISE have reinforced Oxiplant commitment to adopting artificial intelligence as a strategic enabler of innovation. Long-term, the company envisions AI playing a central role in improving quality, efficiency, and competitiveness across its manufacturing operations while supporting its ongoing digital transformation journey.