Quality Forecasting with Signal Processing

Is my production process (still) running smoothly or do I (already) need to intervene manually? These are typical questions that are posed daily in the management of modern production processes. To answer them, up-to-date information is required, which is generally acquired by sensors directly during the manufacturing process.

The department “Machine Vision and Signal Processing” develops monitoring systems capable of detecting errors at an early stage even in complicated situations. There are many reasons why a situation can be highly complicated: the length of time required to assess the quality of a component is often greater than the actual cycle time itself, making it very difficult to detect errors. If a component can only be inspected using a destructive method, inspections have to be carried out randomly. To optimally use available sensor data even in such cases, digital signal analysis and machine learning are implemented.

The monitoring system has a broad range of use, ranging from cyclical productions in the field of plastic injection-molded parts, through ultrasound welding, right up to continuous processes in process engineering.

We also help you select the right sensors and integrate them into your production facility as well as identifying a suitable operating point when they are put into service.


Plastic Injection Molding

The quality of injection-molding processes can be monitored inline; this enables the quality of parts to be assessed by means of fluctuations in signal datasets from the process. The signal datasets are acquired by sensors integrated into the manufacturing tools and evaluated with special one-or two-step classifiers.


Ultrasound Welding

Inline quality monitoring uses sensor data to identify fluctuations in ultrasound welding processes. In addition to the sensors of the welding machine, further sensors (e.g. for measuring force, temperature and structure-borne noise) are also integrated near the joining zones.


Functional Surfaces

A particular feature of the inline quality monitoring method developed is that is can also be implemented to create functional surfaces with micro- or nano-sized structures for use in the manufacture of optical, medical or diagnostic products.


Faster Start-Up Times

A particular feature of the inline quality monitoring system developed is its ability to be rapidly implemented at the start of production. This is based on an initially-reduced number of sample datasets, which is extended as unknown process states are identified.


Energy Monitoring

When optimizing production processes - apart from product quality - more and more emphasis is being placed on reducing energy requirements. Consequently, the modular process monitoring system that has been developed also contains components that record and analyze energy consumption during production.