Quality control innovative project for the Aluminium Rolling industry Auto-Coil Release Assistant solution

Quality control innovative project for the Aluminium Rolling industry

Auto-Coil Release Assistant solution – ACRA is an AI-powered solution specifically designed to enhance Quality Assurance focused on coil surface sorting. This innovative system offers a semi-supervised approach, ensuring that a human is actively involved in the AI-driven process for increased oversight and precision. By automatically sorting each coil based on surface characteristics—such as defects, patterns, and density—ACRA supports more accurate, efficient, and reliable quality assessment while maintaining a vital human touch for decision-making.

BASIC DESCRIPTION

Introducing the Auto-Coil Release Assistant (ACRA) – Auto-Coil Release Assistant solution aims to enhance Quality Assurance procedures related to the surface within coil manufacturing industries. ACRA automates sorting of each coil based on detailed surface characteristics, including defect types, patterns, density, and localization. This advanced system elevates material quality assessment, preserving value and optimizing downstream processing.
The innovative project “Auto-Coil Release Assistant” is designed and implemented in collaboration with Elval, the alumimiun rolling division of ElvalHalcor SA, in the context of Empowering SMEs call for proposals from EIT-Manufacturing.

THE CHALLENGE/THE PROBLEM

In coil manufacturing, existing surface inspection systems, regardless of the technology used, focus on information and metadata of defects’ appearance on the surface of coils. However, the final quality decision remains a manual process.
Skilled operators must rely on their experience and judgment to classify each coil and determine its downstream processing path, often involving multiple departments and considerations. This process is time-intensive and influenced by external business criteria, which can complicate decision-making.
The reliance on manual classification may result in inefficiencies, errors, misclassifications, and excessive resource consumption.

THE SOLUTION

The Auto-Coil Release Assistant solution envisions the modelling and automation of manual processes related to surface quality for coil manufacturing.
AI-driven algorithms aim to automate these non-deterministic processes, while keeping human-in–the-loop. The solution incorporates material-specific, real-time sorting based on AI models, tailored to enhance accuracy across a spectrum of alloys commonly used in coil manufacturing. It is focused on combining existing defect detection results and analyzing patterns produced by different existing machine vision systems, in various formats, with other business-oriented criteria. Additional business criteria that may affect sorting can be modified by the end-user at any time thus keeping the human-in-the-loop.

THE IMPLEMENTATION METHODS

The main solution:

Applies AI-driven algorithms specifically designed for data analysis, pattern recognition, and surface sorting. The AI model is trained on a vast dataset to ensure accurate and reliable sorting recommendations
Integrate a semi-supervised approach, where human operators review and validate AI-driven proposals to refine accuracy and maintain control over quality decisions
Validated in the operational environment of the end-user of the project, focusing on different productions (coils) and client requirements.

THE TAKEAWAYS

Automated sorting proposals

Intelligent data-driven recommendations

Advanced pattern recognition

Human-centered Artificial Intelligence

Business Intelligence

Machine learning algorithms

Human in the loop

Industry 5.0

Innovative quality control solution