The ALFA project fosters synergies between model-based and ML-based diagnostic methods for fault detection and diagnosis in building automation. One objective is to mitigate the so-called "cold start problem" (generally, a lack of data for newly commissioned buildings). To this end, simulations are used to generate an ML model.

The semantic description (ontology) of data points is central to scaling data-driven optimisation measures in buildings. The overarching project goal is the automated acquisition and generation of metadata for building ontologies from the building automation system, using innovative AI-based approaches.

BEYOND aims to develop the technological foundation for "Next Generation Energy Services", enabled by the interplay of the following technologies: virtual reality for visualisation and real-time interaction; machine learning and physical simulations; and IoT platforms for bidirectional real-time communication between buildings and their users.

The BOSS project develops causal AI methods that derive semantic data from time series on the one hand and are applied to FDD use cases on the other. Causal AI offers the advantage of targeting cause-and-effect relationships rather than merely statistical correlations.

The COOL-KIT project develops, demonstrates and structures system solutions for cooling buildings, with a focus on Gründerzeit-era architecture. Selected system configurations are implemented in several buildings of the participating universities, predictive control approaches are tested using a digital twin based on an IoT platform, and everything is comprehensively evaluated in terms of energy performance, comfort, economics and environmental impact.

The exploratory project DomLearn aims, for the first time, to evaluate the potential of domain-informed machine learning for intelligent energy systems together with prospective users from industry. Concrete implementations are also discussed with international experts. Selected solutions are implemented, tested and evaluated in a proof of concept.

Prosumers and energy communities are integral to modern energy systems. By applying machine learning to fault detection and diagnosis, generation and consumption data are used to improve the availability and reliability of technical installations at prosumer level. These efforts are demonstrated and validated at the five international ECom4Future pilot plants and laboratories.

The GreenHeat project develops interpretable AI methods for fault detection and diagnosis as well as for the optimal control of heat pumps. The interdisciplinary GreenHeat project pursues scientific advances that go beyond the international state of the art, while its industry partners aim for global technology leadership in innovative energy services.

Keeping water and energy use in TU Graz buildings continuously in view — the project demonstrates how an Internet of Things platform enables real-time communication between buildings, various systems and staff, while an overarching artificial intelligence optimises and monitors energy and water consumption. The project contributes to the development of intelligent and sustainable buildings and energy systems. INFRAMONITOR is currently being developed further in several national and European projects.

Climate change is producing considerable social and health-related effects worldwide. Heatwaves make everyday life particularly challenging for vulnerable groups. The project's goal is to conduct a comprehensive study in nursing homes and hospitals on contactless vital sign monitoring using radar technology, and on linking occupancy information from the radar sensor to the optimal, energy-efficient control of building services for a healthy indoor climate.

The OctoAI project develops the next generation of high-performance edge AI for intelligent buildings. We combine the concept of edge AI with user-centred energy services and trial two edge-ready applications.

The developments in PersonAI target a radical innovation in user-centred energy services in the building sector through the application of "personal comfort models" (PCM). In a proof of concept, personal comfort models are to be integrated into building automation together with innovative energy services for the first time, with a focus on energy efficiency impacts alongside modelling-specific performance indicators.

In the RINGs project, an optimisation model is developed to evaluate different operating modes of a building complex under varying weightings of cost, self-sufficiency and emissions. From this, rules and strategies are derived to maximise the efficiency of the hybrid energy system with respect to a range of objective functions.

The SAGE project develops scalable multi-agent architectures that enable buildings to detect operational anomalies autonomously and respond dynamically to environmental changes. Integrating multi-agent architectures with large language models (LLMs) and developing a human-in-the-loop approach optimises collaboration between people and machines.

The SELF²B project concerns an AI-based, self-learning and self-diagnosing fault detection and diagnosis (FDD) solution for the building portfolio of Bundesimmobiliengesellschaft. The resulting solutions are demonstrated as a real-time online FDD prototype in live operation (including continuous monitoring of HVAC systems and a PV installation). The project also produces a technology concept for "self-learning, self-optimising" existing buildings for the next generation of efficient building operation.

The overarching goal of the UrbanHP project is a structured analysis of how to integrate heat pumps into existing urban districts in an optimised and grid-supportive way. The analysis draws on detailed energy models, economic and environmental assessments, and a semi-virtual implementation at the real-world case study of a large building complex.

In the VENTUS project, we apply current research in physics-informed AI and probabilistic causal AI to radically optimise the operation and maintenance of wind turbines. Building on an analysis of fault cases and performance degradation, the project pursues an explainable AI system with the potential to reduce losses from downtime and maintenance by 50%.

The central project outcome is a virtual reality digital twin environment of the "My Smart City Graz" and "TU Graz – Innovation District Inffeld" test areas. Within it, users can interactively operate and visualise building energy simulations as well as Internet of Things monitoring data from the districts. The project also answers several research questions in areas not previously addressed.

WELLFIT investigates sleep quality, thermal comfort and productivity using smartwatch-based mobile sensing combined with ecological momentary assessment (EMA). Over the course of the project, nudging and just-in-time adaptive interventions (JITAIs) are developed and tested together with low-tech, low-cost cooling solutions.

WhichWay aims to systematically analyse and compare IoT platforms for the digitalisation of the energy system. Together with international experts, both functional and non-functional requirements for IoT platforms are defined from the perspective of various stakeholders.