AI on the Factory Floor

How we automated the insulation of Electrical Cables

Machine Learning

Industrial

A global leader in automotive cables wanted to optimize its insulation cable process, which relied on manual checks. We built and deployed an ML model on factory floor machines, reducing waste, saving €300 per day per machine, and freeing operators for other tasks.

Implementation time:

5-7 months

daily savings per machine

300€

insulation material saved per run

12k

extrusion recipes modeled

+ 1k

The Opportunity

The company understood they had to optimize its isolation cable process, because:

Manual Process

Quality checks relied on human intervention, increasing the risk of error during production

‍

High Material Waste

Delays in decision-making caused unnecessary losses of copper and insulation material

‍

Variable standards

The start-up phase of extrusion differed by recipe, making it harder to ensure consistent product quality

The Solution

We deployed a machine learning quality decision model based on historical data that runs on factory floor machines. In real time, it determines when the cable is approaching target properties.

‍

A key aspect of this solution was inference time. Each second without a decision meant costly material waste, so predictions needed to be immediate. Our model delivers real-time predictions, using only the minimum material required.

‍

We deployed our machine learning model within the factory floor machines, ensuring seamless integration with existing systems. This enabled the client to leverage the benefits of AI on legacy machines without incurring costly infrastructure upgrades.

Meet the team of this project

Fábio Cruz

Senior Machine Learning Engineer

João Gonçalves

Machine Learning Engineer

The Impact

The results were significant. Each machine now saves over €300 per day, representing million-euro annual savings across the client’s operations. More than 1,000 extrusion recipes are modeled, and 12m of copper and insulation material are saved on every run.

‍

By automating quality verification, operators are freed from constant monitoring and can focus on higher-value tasks. The client was thrilled with both ROI and operational efficiency, and based on this success, we are now expanding the solution to improve their cable design process. This is fully aligned with one of DareData’s main goals: delivering constant value via innovation while building long-term strategic partnerships.