The OptiNETS project was designed to investigate, develop and make available methods and tools that enable the effective and efficient analysis and optimisation of energy systems and networks through the use of artificial intelligence (AI) and machine‑learning (ML) techniques at a virtual level. The focus was on application scenarios and research questions that arise in the context of renewable energy and the energy transition. By addressing the shift from centrally controlled to decentralised generation, from demand‑driven to supply‑driven policy, and towards a more democratic energy supply, the project sought to harness AI/ML solutions for systematic, goal‑oriented analysis and optimisation of energy systems, thereby advancing the field of Cyber‑Physical Energy Systems (CPES).

At the outset, AI and ML had already proven highly successful in domains such as speech recognition, translation, and strategy games, where problems are characterised by high complexity, heterogeneity, and dynamic conditions. However, their application to the planning, analysis, optimisation and operation of energy systems had been limited to isolated cases, for example in plant control or weather and market data simulation. OptiNETS therefore aimed to bridge this gap by building on a pre‑existing methodology and tool environment for modelling, simulation and analysis of energy system scenarios.

The project was organised into three work packages. Work package 1 focused on research into innovative AI/ML approaches for planning and operation of energy systems. Work package 2 translated these research findings into practical solution methods and integrated them into a usable tool environment. Work package 3 applied the developed methods to concrete scenarios provided by the partner utilities Gridsystronic Energy and Stadtwerke Ludwigsburg‑Kornwestheim. The schedule, budget and cost plan were largely adhered to, and all planned tasks were completed, with the project even exceeding its original expectations.

Key scientific outcomes emerged from the research. In the first application case, a range of ML approaches and forecasting strategies were explored to build accurate and efficient simulation and prediction models for energy plants. Convolutional Neural Networks and Temporal Convolutional Networks achieved average prediction accuracies of up to 98 % for hourly forecasts and 91 % for day‑ahead forecasts, which are relevant for intraday and day‑ahead electricity market trading. The second case demonstrated an AI‑based optimisation of plant and plant‑cluster operation schedules tailored to specific objectives such as residential‑tariff models, market trading, or CO₂ emissions. Compared with conventional operation, the AI‑driven schedules improved performance by up to 24 %. The third case addressed technology‑mix optimisation for the expansion or contraction of plant clusters, optimising the sizing of technologies such as photovoltaic, wind, combined heat and power, and others. The fourth case integrated storage technologies into the planning and operation of micro‑grid scenarios, further enhancing flexibility and resilience. All four solution methods were incorporated into the project’s tool environment, making them accessible for practical use.

The project ran from 1 September 2018 to 31 August 2022 and was funded by the German Federal Ministry of Education and Research under grant number 3FH7571X6, within the “Engineering Talent – Cooperative PhD” programme. The project was led by Prof. Dr. Joachim Gerlach of Hochschule Albstadt‑Sigmaringen. Collaboration with the partner utilities Gridsystronic Energy and Stadtwerke Ludwigsburg‑Kornwestheim provided real‑world application scenarios and facilitated the translation of research into practice. The resulting tool environment, together with the demonstrated performance gains, represents a significant contribution to the optimisation of sustainable energy systems and offers a foundation for future research and deployment in the evolving energy landscape.