How SMBs are making production smarter and more efficient with data

The Dutch manufacturing industry faces a double challenge: customers expect more flexibility and customization, while staff shortages, rising costs and tight margins put pressure on production capacity. At the same time, machine learning (ML) offers concrete opportunities to produce more efficiently, reduce errors and make better decisions based on data.

‍

In this white paper, we highlight two hands-on applications of ML that are directly relevant to SME manufacturing companies: visual quality control and process optimization.

‍‍

Application 1: automatic quality control with computer vision

‍

Situation
A plastic injection molder in Almelo produces technical parts for the automotive sector. The quality requirements are high: even minimal distortion or discoloration can lead to customer rejection. Until recently, employees manually inspected the parts for visible defects. This regularly led to missed defects, resulting in returns and costs.

‍

Solution
The company installed a camera system at injection molding machine outlets and coupled it with a machine learning algorithm trained on thousands of sample images of correct and incorrect products.

‍

How it works

• During production, the camera continuously takes images of each part
• The algorithm compares these with the ideal profile and detects deviations such as air inclusions, burrs and discoloration
• Erroneous parts are automatically removed from the line

‍

Result

• 95% of errors are now automatically recognized
• The number of returns decreased by 60%
• Customer satisfaction and delivery reliability increased noticeably
• Operators can focus on the process instead of visual inspection

‍

This application directly improved quality, reduced costs and increased output per FTE.

‍‍

Application 2: process optimization through data analysis

‍

Situation
A metal processing company in North Brabant produces parts in small batches on three production lines. Planning seemed efficient, but the production manager suspected that lead times were higher than necessary. There was no clear data on where in the process delays were occurring.

‍

Solution
Working with a data specialist, the company gathered data from the PLC systems, the ERP package and through manual input from operators. A machine learning model was then trained to identify bottlenecks in the process.

‍

 What transpired?

• Setup times varied widely between shifts, especially for more complex jobs
• Shift transfer resulted in an average of 20 minutes of delay per shift
• One line was running structurally less efficiently due to improperly planned series sizes

‍

Actions

• Setup procedures are standardized based on best practices from the model
• The shift transfer was adapted with a digital transfer form
• Production planning was optimized based on data analysis

‍

 Result

• Average turnaround time decreased by 14%
• Output per shift increased 11% with no additional hours
• Operators are more engaged thanks to insight into their performance

‍

This project shows that even simple data insights, combined with ML, can lead to tangible improvements in efficiency and engagement.

‍

Conclusion

Machine learning is not a distant vision of the future. SME manufacturing companies in the Netherlands are already using it to improve quality and streamline production processes. Visual inspection and process optimization are two low-threshold applications that deliver quick results if applied properly.

‍

TMC Media helps manufacturing companies translate this technology into clear strategy and communication. Whether you are already collecting data or just looking for a practical start: we help you make the step to smart production concrete. 

‍