Digital twins for Operational Management Systems (OMS) and production analytics

All manufacturers, be that industrial, in energy, health, fish farms or any other, are in need of systems to manage their processes. Systems that allows you to evaluate how changes, small or big, influences the outcome by tracking the process and keeping the data. Why did the bread that left the bakery that morning have such a crispy crust? At what time do you achieve a maximum cost/benefit ratio in regards to cleaning the photovoltaic panels in your solar farm? Why did todays batch of soap end up bad? Can it be tracked to a faulty raw material? The cost of both purchasing and implementing systems that gives you the level of control needed are generally very high. For the typical SMB this means that you'll have to make do with a manual process, adding to the already high workload. Larger enterprises spending thousands of man-hours configuring and maintaining such systems are common. SkyTrack is a project that aims to bring down the barriers that holds back a wider adaptation of process tracking software: - Through the use of cloud based architecture and existing software will reduce the cost of installing and maintaining. - By implementing generalized and automated models will minimize the need for man-hours. - With extensive use of digital twins will allow these processes to be communicated with, perform calculations and other semiautomatic actions. - By focusing on user experience from day one, will bring down the need for specialists. - Industry standards, such as OPC UA, will prevent vendor lock-in.

H1:Information modelling Det har vært jobbet med flere informasjonsmodeller innen fornybarindustrien (solkraft, vindkraft, vannkraft) i OPC UA. Disse er i bruk hos kunde og utvides stadig. Det er utviklet metoder for å utvide informasjonsmodeller slik at de kan inneholde direkte kalkulasjoner slik at manuelle prosesser elimineres. H2:Semantic mapping and name-space transformation Det har blitt gjort flere forsøk på å etablere gode metoder for semantisk mapping, blant annet ved utvikling av egne verktøy. Selv om noe av dette er i bruk, har hovedretningen rundt mapping i større grad blitt basert på malsett og gode integrasjonsverktøy. H3:Automatic digital twin generation Arbeidet er fordelt på to hovedretninger: Burk av eksterne rammeverk for automatisk opprettelse av digitale tvillinger på den ene siden, samt kalkulasjonsrammeverk basert på informasjonsmodel som den andre. Det første prosjektet, avsluttet i begynnelsen av 2021, dannet grunnlaget for mange av de arkitekturvalgene som nå tas i bruk i vårt hovedprodukt. Det andre er en domenespesifikk (solkraft) løsninger som tillater en tilnærming til digitale tvillinger basert på OPC UA (denne er i bruk). H4:Architecture and patterns Flere arkitektur- og mønsterprinsipper er evaluert og tatt i bruk; mikrotjenestearkitektur, strømmende data, «actor pattern», distribuerte systemer for nær real-tid, NOSQL og graph-databaser (for å nevne noen). Flere av disse har enten blitt eller skal bli en del av vårt hovedprodukt. H5:User experience (UX) optimization Bakenforliggende strategi for utvikling av brukergrensesnitt er byttet (mikrotjenester og React, OPC UA tilgjengelig som REST og med egen indeks for raske oppslag). Flere grensesnitt utviklet, og disse oppdateres automatisk basert på informasjonsmodellen (autogenerert innhold). Arbeid med å lage et eget REST-API basert på digitale tvillinger er underveis (TWIN-API).

Production companies need systems for execution, reporting, analyzing and optimizing their production. These systems may be computer-based OMS systems - meaning Operation Management Systems, or they are ad hoc systems based on paper, Excel sheets etc. The project focuses on making applications and tools that dramatically lowers the barriers of cost and complexity for realizing and operating such systems. The target system based on the project findings will be cloud hosted and OMS- and related applications can be generated automatically from generic plant information models (digital twin patterns). One OMS related field which is presently often neglected, is the preparation of data, prior for applying artificial intelligence (AI) and machine learning (ML) methods. A lot of companies and methods has been created for AI/ML purposes. The methods planned to be developed in this project is almost a prerequisite for the success of such applications. R&D challanges targeted in the SkyTrack project are: - To classify plants into types that can be represented by similar digital twins. - To find algorithms and methods for automatic or semiautomatic transformation of signal names (tags) and parameter names in plant databases into SkyTrack digital twins names. - To find methods for automatic generation of plant models into digital twins. - How to optimize the user interface to enable end users and system integrators to configure and realize cloud based OMS based on basic information models and automatic application generators. Solving these research challenges will make possible to reduce the engineering side of the cost of realization of an OMS applications. In addition, the entry cost will be small and scalable for the user.