Pre-engineering as the key to success

A well-founded analysis of the customer's needs creates the basis for an individual solution for visual inspection. For this purpose, a manual inspection is carried out in the company's own laboratory based on the customer-specific test criteria and the corresponding test sets, consisting of good and bad samples. The various errors are then verified and documented. From this, a common understanding of inspection tasks is derived and defined together with the customer. The holistic analysis also includes other test methods for leak testing, Headspace Analysis (HSA) and Near Infrared Spectroscopy (NIR). A combination of various inspection technologies with camera- and X-ray-based visual inspection can significantly increase process reliability. For example, the combination with X-ray technology can detect embedded glass particles in Lyo cakes and leads to an additional increase in the detection rate of up to 12%. Likewise, a combination with 100% NIR spectroscopy for the detection of defects of lyophilizates (meltbacks, etc.) can significantly improve. The result is a tailor-made inspection solution tailored to customer requirements.

Customer-specific system design

Based on the knowledge gained from pre-engineering, the required hardware components are defined and a customer-specific optical design is created. This ensures that components are used which correspond to the latest state of the art. This is followed by images of image sequences and the collection of image databases in the laboratory. On this basis, the evaluation of the most powerful inspection algorithms for each individual test station is carried out. Both the software and the hardware can be customized to a high degree. For each inspection task there is a wide range of different algorithms, cameras and lighting available, the combination of which is created individually.

Individual recipe development

The image data is analyzed and evaluated in the further process. Based on a large number of pre-configured tools, the parameters are optimized to increase the test performance (False Reject Rate and Detection Rate). Deep learning algorithms are also increasingly being used. These were specially developed to improve the distinction between air bubbles and particles, for example. This results in customer-specific recipes. In this phase the optical equipment such as camera position, light adjustment, hardware components etc. are completed. In the machine itself, the optical path remains static. Since there are no moving parts such as mirrors or cameras, the susceptibility to errors is reduced to a minimum and the service life of the components is increased. Furthermore, the optical path is developed in such a way that the entire format range is always covered and therefore no further adjustments in the optical structure are necessary, which can also lead to error results.

Fine tuning

The automated visual inspection solution is defined on the basis of the test results and customer requirements. The recommendations of the optical design and the software algorithms are documented for the customer. With the help of the integrated offline mode, a fine tuning of various shape parameters is carried out and new shape parameters are generated. Thanks to the flexible alignment and the modular design, a simple transfer from the laboratory to the production system is possible.

The modular system has a test rotor with 60 servo controlled stations which achieves a performance of up to 600 specimens per minute. Up to 13 cameras inspect a wide variety of containers. The machine design is designed so that additional testing technologies can be integrated, re-inspection and validation runs can automatically run through the machine again, and individual discontinuation modules can be configured.

WILCO AG has developed a system called VIVA (Visual Inspection Virtual Assistant - a virtual assistant for visual inspection) for the re-inspection of grey test specimens. Here, images are made available to the operator for re-inspection via a cloud or server. The required evaluation by an expert can then be carried out independently of location on the basis of high-resolution images on the screen. The support of VIVA enables a close-up analysis and the image sequences can be repeated. This can be used to create a database for gray and bad objects and classify them accordingly.

Qualification of the system

The WILCO Vision team is available to provide support during the qualification of the system. Function tests and qualification tests are carried out. Support for the scarcity test is also part of the service offering. Comprehensive training is taken for granted.

Further use is ensured in a sustainable manner

Thanks to the use of existing inspection functions and the modular system design, upgrading to new formats is very easy. The WILCO vision team supports the implementation of new inspection functions or the extension to new formats and products.