AIOptiGrid unites multidisciplinary data analytics and artificial intelligence to turn Croatia’s electricity grid from a reactive, siloed system into a proactive, predictive one — built by PMF, FER and ADNET, and validated by HEP ODS.

36

Months
duration

€2.98M

Total project
value

€2.57M

Grant
amount

4

Consortium
partners

Europe’s energy transition is straining electricity distribution systems: mass electrification and the rise of renewables, storage and prosumers demand networks that are more predictive than today’s largely siloed SCADA, GIS and smart‑metering tools allow. The European Commission estimates €584 billion in grid investment is needed by 2030 – with limited time and people to plan it.

AIOptiGrid responds with a cross‑disciplinary, AI‑powered platform that correlates technical, meteorological, socio‑economic and environmental data to reveal dependencies invisible to conventional planning — such as how weather, urbanisation or consumer behaviour shape grid performance.

Piloted with HEP ODS, Croatia’s national distribution operator, the result is faster renewable connections, shorter outages, optimised maintenance and investment planning that is transferable across EU DSO frameworks.

The project delivers an AI‑powered software platform designed to be adaptable across regulatory and technical contexts — deployable cloud‑based or on‑premise by other DSOs.

LAYER 01

Central Data Warehouse

A unified data lake integrating SCADA, GIS, smart‑metering and external sources — ensuring interoperability and scalability across the grid.

LAYER 02

Predictive Middleware Engine

AI and statistical models performing correlation analysis, anomaly detection and scenario simulation across heterogeneous datasets.

LAYER 03

UI & Visualization Layer

Interactive dashboards, financial modules and configurable tools supporting planning, maintenance and day‑to‑day decision‑making.

Unlike commercial smart‑grid tools, AIOptiGrid fuses non‑technical signals with grid telemetry – including AI methods adapted from natural‑language processing — to surface dependencies no single dataset reveals alone.

  • Technical & operational grid data
    SCADA, GIS and smart‑metering telemetry from HEP ODS.
  • Meteorological & climatological variables
    Weather patterns and extreme‑condition exposure.
  • Socio‑economic & demographic indicators
    Urbanisation, population and consumer‑behaviour data.
  • Environmental & market data
    Emissions context and energy‑market signals.

A 36‑month work plan led by academia, engineered with industry, and proven on Croatia’s national distribution network.

Lead partner

PMF

Faculty of Science, University of Zagreb — project coordination, AI model design, mathematical analytics, data preprocessing and research validation.

Technical partner

FER

Faculty of Electrical Engineering and Computing — technical modelling, forecasting and optimisation algorithms for power systems; scientific evaluation.

Industry partner

ADNET

Software architecture, system integration, interface design and commercialisation through the NETVISION platform.

End user

HEP ODS

Croatia’s national DSO — end‑user validation, data provision, pilot‑site coordination and operational feedback.

  • Establish project governance and infrastructure
  • Integrate and preprocess datasets
  • Develop and test initial AI models

Milestone — Central Data Warehouse operational; baseline models validated.

  • Complete AI model training and optimisation
  • Build prototype software platform
  • Begin pilot testing with HEP ODS

Milestone — First functional prototype and field validation achieved.

  • Final validation, impact analysis and scaling roadmap
  • Technology transfer to ADNET and deployment at HEP ODS
  • Dissemination of results and scientific publication

Milestone — System ready for commercialisation and replication (TRL 7–8).

Predstavljen projekt AIOptiGridNews

Predstavljen projekt AIOptiGrid