Model driven engineering of operations management systems (OMS) using standardized plant and process models

The goal of the project has been to use digital models of industrial plants to support developing and scaling software for these assets. To further this goal, a more effective, interactive way of integrating different sources of digital models in a representation called a graph has been developed. The project has then considered how these graphs can be used to develop analytical software and to develop software for tracking objects in production. The project has developed a new methodology and tool for contextualized access to time series that supports a combination of cloud- and locally based infrastructure. The method works in situations where existing tools do not, and has far higher performance in situations where existing tools work. The project has also resulted in a new approach to building software for object tracking that can make it easier to develop and maintain this kind of software, since the software can reconfigure itself when the asset and the corresponding digital models are updated.

The results can potentially simplify the development, maintenance and scaling of industrial software. The project has led to new knowledge within the use of knowledge graphs in industry through a new methodology that has been shown to solve problems that have not been able to be solved so far in a way that has better performance than existing approaches. Due to the fact that the methodology is largely based on open, established standards, it can potentially represent an open ecosystem for industrial contextualization that has greater competition, more innovation and better conditions for SMEs than the existing closed platforms. The project has given Prediktor expertise in graph technology and model-based software development, which now forms the basis for the new software in the investment area of asset management for renewable energy (solar, wind). The principle of using queries to integrate software is used in this software.

Companies managing physical processes need to supervise the operation of the processes involved. E.g. in a bakery one might want to know how recipe- and operating parameters, such as baking temperatures and time, influences the produced bread parameters such as taste, crust, consistency and looks. Many producers use manual systems to register production data based on e.g. paper or in Excel. Big companies often invest in computer-based systems, often termed Operation Management Systems (OMS). Realization of a computer-based OMS is usually expensive, due to the amount of labor required to build a new system. The goal of this project is to find methods so that OMS applications can be developed with minimum engineering time by persons holding limited knowledge of the software, but with knowledge about the actual production facility and its needs. Model driven (software) engineering is about doing software development by manipulating domain specific models instead of writing code. Code is in turn generated automatically from domain specific models in prescribed ways. This PhD will address methods for minimizing the manual effort needed to construct an OMS by using a model-driven approach that leverages the standardized information sources existing in a production unit. The basic idea is that these information sources can be understood to partially specify OMS software, and that engineering time can be further reduced by integrating these data sources iwhen specifying the model. This research problem is the first among the open research problems listed by Vyatkin (2013) in his review of Software Engineering in Industrial Automation.