to all Blog Posts

AI and ML helps companies become more productive

Published: · Last updated: · 3 min reading time

How do you distinguish semantic models in principle? Are there groups/types of models, or what are the most important criteria?

Olga Mordvinova:Models based on semantic information do not only consider measurement data such as those supplied by machines or individual sensors but also consider business processes, relationships, and meanings of the various data structures. This makes them much more effective in their evaluation of relationships, correlations, and ultimately in the calculated predictions.

What makes the semantic model of FORCAM special compared to others?

Olga Mordvinova: FORCAM’s semantic model is based on a profound knowledge of the manufacturing process, its versatility, and pronounced variability.

It exposes in its interfaces pre-interpreted production data through consistent data structures, which provide not solely machine or workplace information, but also detailed information about the historical and current state of the cross-job production process.

Wouldn’t it make more sense for all providers to use a uniform semantic model together instead of each using its own? The advantage lies in the mutual understanding (uniform language) and not in the differentiation through a particular language/semantics.

Olga Mordvinova: Standardization is always desirable – the keywords being “scalability,” “platform economy.” A uniform model could, for example, help the industry to exchange information efficiently and openly between applications and programs, simplify the integration of data sources and functionalities, and avoid data silos, which are often a limiting factor for successful digitization.

The readiness for standardization is undoubtedly one of the components that make the FORCAM semantic model so successful and influential. Extensive and open interfaces (OPEN API), as well as a bi-directional integration of systems relevant in production, are the foundation of FORCAM’s openness and harmonization ability. Whether MES, ERP systems, or even the machine connection, this data is semantically interpreted and made available for broad, open use.

However, the standardization of data structures and the openness of controlled data provision are not FORCAM’s only strengths. Just as important is the full understanding and in-depth support of flexible manufacturing processes. The mastery of complex interrelationships is possible in connection with the FORCAM system, whether in the process sequences or the reporting area (e.g. how to calculate an OEE key figure correctly). In my opinion, it provides a necessary and reliable basis for all applications that are based on production data and represents a real added value for the employees in the production.

Which semantic models do other platform providers use, and what are the main differences?

Olga Mordvinova: Most IoT platform vendors today focus on generic approaches and approach technical integration of machine data or external systems. Modeling and providing the information needed by employees in production is often complicated and is often “modeled in addition” not by platform vendors themselves but by applications on the platform.

Which semantic models do other platform providers use, and what are the main differences?

Olga Mordvinova: Most IoT platform vendors today focus on generic approaches and approach technical integration of machine data or external systems. Modeling and providing the information needed by employees in production is often complicated and is often “modeled in addition” not by platform vendors themselves but by applications on the platform.

For which use cases (and for which customers) do you already use AI in combination with FORCAM solutions today?

Olga Mordvinova: The AI applications of incontext.technology GmbH master the semantics of the production processes of FORCAM FORCE and learn to automatically recognize problematic situations even from the “small data” of the production line. It combines self-learning algorithms with a global insight into the production and the ability to provide local instructions at workstations to the employees, depending on their role. Examples of our application are predictive maintenance of workstations, detection of time-critical orders, or predictive overall line effectiveness.

What do you see as the next essential fields of application for AI in the industry?

Olga Mordvinova: Whether it is in maintenance or quality inspection, we will continue to move further and even more towards process automation. While the machines continue to perform repetitive tasks, and AI continues to optimize operations, humans will have time to deal with complex tasks.

Interview: Victor Gruber