AI-enabled quality assurance
To improve quality in a sustainable way!
There are different quality requirements within industrial production chains. To meet these, companies tend to rely on manual quality checks. However, these are generally error-prone, burdensome and expensive. AI solutions enable real-time quality control to be automated, reliable and logged in detail. This makes it possible to identify and prevent errors in production and logistics quickly and effectively.
State of play
One of the biggest topics in the area of AI is image recognition. This, in turn, is one of the key enablers for AI-enabled quality assurance. Developments in this area have been very advanced in recent years. As a result of the high volume of investment, especially large firms, both market growth and the application spectrum have steadily increased.
The use of automated quality assurance systems is also becoming more attractive and economical for small and medium-sized enterprises.
Technology and deployment
Description of technology
Camera systems involved in the production chain provide real-time images of the products being manufactured/processed. AI allows automated image processing software to detect product-specific fault images (e.g. cracks, missprays, supernatants, geometric deviations). To this end, complex artificial neural networks are trained in order to meet the respective production standards, using many illustrative pictures tailored to each quality control and their individual specifications.
AI algorithms also make it possible to identify the smallest errors, which are often overlooked in the manual quality control, with permanently reliable quality. The resulting quality reports can be used to establish correlations between, for example, the adjustment of machinery in the production process, environmental parameters such as pressure, temperature or similar and the resulting quality, thereby sustainably improving the production process.
Possible use scenarios
Gradual introduction
Step 1: Analysis of the actual situation
— Which class of problem is relevant in your quality assurance? (Cracks, fitting defects, soiling, breaking points, text testing, etc.)
— What material are your manufactured components?
— Is it possible to integrate cameras in the current process?
— At what interval do objects on a production line have to be tested?
Step 2: Data acquisition
If the installations are not capable of generating corresponding image data, an upgrade may need to take place upstream.
Step 3: Interchange
The corresponding image processing software must also be selected or implemented.
Step four: Usage
On the basis of the collected data, overall product quality reports can be produced and frequent sources of errors in the production process can be identified, analysed and rectified iteratively. A continuous process of improvement can therefore be initiated.
Opportunities for SMEs
contact
Use our technology radar to keep a look at the main technologies relevant to SMEs!