Data obtained in the automatic inspection process offer enormous potential. One of the major challenges of the future will be to use it even more intelligently. We always speak of Smart Data when it can be used for automation and process control on the basis of intelligent algorithms. VITRONIC pays special attention to this in its product development. For example, we are working on the use of deep learning methods for individual inspection applications in order to further improve the processes and results in error detection. Another concrete field of application is the process-related visualization of the collected inspection data in the form of heat maps and trend analyses. These can be used in a variety of ways, for example to optimize process parameters or to determine the changeover time for critical spare parts.
Standardized interfaces are an important basis for the exchange of this information. To ensure that the systems can communicate in one language in the future, VITRONIC actively supports the OPC UA Machine Vision Initiative. The goal is to simplify data exchange and networking on the automation level which enables a more flexible use of vision systems. Accordingly, an M2M interface based on the "OPC UA for Machine Vision" Companion Standard will be available with the next release of our basic software framework (Q2 2020) as a standard module for all VITRONIC automation solutions.
Machine vision becomes machine learning
A promising development topic is the so-called “Closed Loop”, i.e. the automatic trend analysis in a closed system. Here, in the automated application, the information from inspection is evaluated, fed back directly into the machine control system, where it can automatically correct machine parameters. In other words: The machine adapts automatically. In the “Fast Feedback Loop”, smaller sub-processes are broken down into sensible sub-units in order to identify necessary adjustments as early as possible and to react to them automatically.
Last but not least, we still see great potential in the area of predictive maintenance. This approach is based on analyzing data and process parameters to predict when a component within the process on the production line will require maintenance or replacement.
These measures enable the manufacturer to significantly increase the output quantity and reduce downtimes at the same time.