The Impacts of Semantic Technologies on Industrial Systems

The Impacts Of Semantic Technologies On Industrial Systems

Industrial systems face the challenges of robust and flexible control of industrial processes while satisfying the demand of mass customization and reduced time to market. To meet these requirements, systems need to work in a distributed manner, because the traditional centralized approaches are not sufficient. However, the distributed architectures used in industrial control systems are still tightly coupled from the point of view of system integration. To achieve automated cooperation and coordination of distributed components at a larger scale, semantic technologies are necessary to enable truly open systems that can communicate while the configuration of the system is dynamically changing.

Today, the systems used in manufacturing industry are programmed with the focus on performing particular tasks rather than on interoperability in a dynamic environment. This is understandable, but in order to achieve better integration from the shop floor level up to the level of virtual enterprises, these systems have to use explicit semantics to describe the interpretation of the data they provide. So far, the interoperability has been resolved mainly at the physical and syntactical level. We can see the parallel with the World Wide Web (WWW), where the level of exchanging and interpreting documents is resolved, but the semantical level is being investigated within Semantic Web research. As the networked and distributed industrial systems have been influenced by the traditional WWW, they are apparently going to be influenced by Semantic Web as well.

In this chapter, we review state of the art of the research of ontologies, Semantic Web and Semantic Web services as the most relevant technologies enabling proliferation of semantics in industrial systems. We pay special attention to possible applications in distributed industrial systems and show how semantics can be applied to such systems. We focus on the multi-agent and holonic techniques that provide suitable paradigm for such systems and provide clear modeling framework for introducing semantics. We also review the ongoing efforts to create general purpose reusable ontologies for the industrial domain.

We discuss specific characteristics and requirements of industrial automation domain in contrast to the common issues of Semantic Web. The Web technologies have to handle the problems of semantical heterogeneity, inconsistency and uncontrolled behavior of individual sites. In the industrial domain, this can be avoided to some degree. The impact of discrepancies is also different – for instance inaccurate search result is acceptable, but inaccuracy in industrial domain may lead to potential damage of equipment, unnecessary material consumption, or delays in delivery of the manufactured product. The important difference is that we are dealing directly with the physical components in a real-world environment.

The applications of semantic technologies in industrial domain are illustrated on two research projects of Rockwell Automation: (i) semantics-enhanced agent-based material handling control system and (ii) structural search.

The first project aims at adding semantics to a multi-agent control system for material handling tasks, like for example transportation of materials and semi-products within an assembly line. We show how semantics can be utilized for reasoning in agent’s knowledge base. The choreography and orchestration of Semantic Web services provide important inspiration – in reconfigurable industrial systems, it is needed to discover suitable service providers, negotiate the contract, monitor task execution and resolve potential runtime problems.

The latter project focuses on structural search. One of the core applications of the Semantic Web is semantic search, i.e., search within semantically enriched data. The design, operation and maintenance of a manufacturing system is very knowledge intensive task and involves handling of information stored in different forms – for example function blocks or ladder diagrams describing the real-time control system, SCADA (Supervisory Control And Data Acquisition) and HMI (Human Machine Interface) views, collected historical events, etc. We show the advantages of the usage of Semantic Web technologies for enabling search within structured and integrated data. The result helps humans to express queries that involve not only looking for keywords, but also structural relations between pieces of information.

Another potential application of semantic-based systems in manufacturing is the area of human-machine interfaces. They could serve for visualization and further analysis of information where the direct linkage of the manufacturing-oriented semantic systems with the Semantic Web might be required.