From March 13th to 14th, 2023, the consortium of the Delfine project met for the fifth consortium meeting at the consortium partner devolo AG in Aachen. The main topic of the event was the ontology-based Semantic Middleware and the implementation of the Delfine Green Power Index (DSI).
On the first day, the meeting was opened by Nadir Pieper from Stadtwerke Trier (SWT) with a review and the current project status as well as an outlook on the next steps. After an overview of the timing of the event, there was an overview of the project schedule and a review of completed work packages and milestones. In addition, some organizational issues were discussed with regard to the remaining project duration.
The team led by Professor Hendro Wicaksono from the Constructor University (CU) then presented the current status of their work. It started with a talk by Tina Boroukhian (CU) on the integration of the ontology into the Delfine Semantic Middleware. The innovations of the data structure and platform solution were presented. Atit Bashyal and Kritkorn Supyen (CU) then presented their results for the respective work packages. Kritkorn Supyen started with a live demonstration of the created Cypher Query Tool, which is used as a natural language interface to Semantic Middleware to query historical and forecast energy data.
Based on this, Atit Bashyal presented the current status of the forecast API for the forecast data and the underlying structure of the applications. For demonstration purposes, the Django web application was demonstrated to query the forecast data and historical data. A current paper from the Constructor University on the subject of artificial intelligence for demand response applications was then presented.
In the next presentation, Daniel Diewald from Stadtwerke Trier presented a study paper from the Trier University of Applied Sciences that was created as part of the research project on the subject of dynamic electricity prices for companies and the Delfine Green Electricity Index (DSI). In it, current electricity price models for companies were compared with the Delfine solution and transferred to the project scenario. The results showed, among other things, that the tariff models with incentive signals, such as the DSI, offer great potential for the application partners. A major obstacle is the missing roll-out of the intelligent metering infrastructure in Germany.
After a short break, the interactive part continued. There was a lively exchange on the various applications and interfaces that are made available and used in the Delfine project. Change requests, suggestions for improvement and questions were discussed and recorded among the partners.
On day 2, the event was opened by Martin Trat from the Research Center for Information Technology (FZI) with a review of day 1. Meanwhile, the consortium was able to discuss open questions from the presentation.
A live demonstration of the status of the visualization application, which was developed for users by Victor Häfner from the Karlsruhe Institute of Technology (KIT), was then presented. The users could ask questions about the application and express requests for improvement. For example, the presentation of key figures in connection with production optimization was discussed.
The next presentation was given by Mischa Ahrens (FZI) on load forecasting and MILP (Mixed Integer Linear Programming)-based production optimization. The innovations presented include an improvement in the accuracy of predictions in the production process through the use of new user training data.
Based on this, his colleague Mine Felder (FZI) presented the current status of reinforcement learning (RL). In this context, it was shown how heuristics were used as a comparison method and how the synthetic extension of the production data can be used in the RL approach.
Manuel de Melo from Pumacy presented the status of the demonstration and validation of the process data. He presented a summary of recommendations for the application partners as well as the web application with a data entry mask, which is used to validate the data. In this context, it was noted that production optimization can be improved by intervals in the seconds range.
Subsequently, Christoph Dorus from StromDAO presented the business model development that could be considered for a possible use of the DSI. For this purpose, Christoph Dorus and Thorsten Zörner (StromDAO) presented examples of current StromDAO projects in the field of dynamic electricity tariffs and their business models.
In the last lecture of the event, the carried out and upcoming dissemination activities and workshops were presented by Martin Trat (FZI). Finally, the consortium agreed on the upcoming tasks.
The next consortium meeting of the Delfine project is expected to take place in June 2023.