The service is intended to be evaluated using historical datasets containing both normal operation and leak scenarios. The evaluation framework would separate training and testing data to ensure a robust assessment of detection and localization performance. Performance is expected to be measured using standard classification metrics, including accuracy, precision, recall, F1-score, and false alarm rate for leak detection, along with localization accuracy for identifying the leak position. Visual analysis of sensor signals and detected events would further support validation of the model’s ability to capture abnormal behaviour.

The Multi-Energy Demand Forecasting service delivers AI-driven forecasts of electricity and hydrogen demand using time-series data from integrated energy systems. Based on advanced deep learning architectures (LSTM, TCN, Transformers), the service models interdependencies between electricity and hydrogen demand across coupled systems such as microgrids, industrial facilities, and energy hubs, supporting operational planning and resource allocation.

The Operational Scheduling for DHN service generates optimal operational schedules for the district heating network by translating demand forecasts and real-time data into actionable control strategies. Using forecast-driven optimisation combined with rule-based and data-driven techniques, the service recommends supply temperature setpoints, load distribution plans, and operation timelines to maximise energy efficiency and minimise operational costs.

The Anomaly Detection and Fault Diagnosis service detects abnormal patterns in the Torrelago district heating network using AI-based techniques applied to real-time and historical data. By establishing expected behavioural baselines from historical patterns and continuously comparing real-time data against these baselines, the service provides early warnings of inefficiencies, faults, and unexpected operational conditions.

The AI-based Grid Topology Identification service provides automated detection and validation of electrical connectivity relationships within distribution networks by analysing correlations in measurement data. By reconstructing the most likely topology from operational data, the service improves digital grid model accuracy and enhances the reliability of state estimation, load flow simulations, and congestion management tools.

The AI-based Load Forecasting service provides short-term electricity demand forecasts for distribution grid assets and aggregated network areas at 15-minute resolution for horizons up to 48 hours ahead. Using machine learning models (gradient boosting, RNNs, LSTMs) trained on historical consumption, weather, and calendar data, the service supports operational planning, congestion management, flexibility scheduling, and state estimation.

The AI-based Power Profile Nowcasting/Forecasting service creates high-fidelity virtual sensors for distribution grid assets lacking real-time metering. It delivers nowcasts (current-moment power estimates) and forecasts (up to 48h ahead) of active (P) and reactive (Q) power for unmetered assets by combining static asset metadata with dynamic weather and temporal inputs. This replaces static Standard Load Profiles with dynamic AI-driven time series that close the observability gap in distribution networks.

The AI-Assisted Predictive Maintenance service transforms distribution asset management from reactive/time-based to condition-based maintenance. By fusing historical operational data, chemical diagnostics (DGA), and high-frequency sensor streams (vibration, thermography), the service calculates a normalised Asset Health Index (AHI) and predicts the Remaining Useful Life (RUL) of critical infrastructure including power transformers and switchgear.

The AI-Phase Identification service determines to which of the three electrical phases (A, B, or C) each smart meter or consumer connection is physically attached in LV distribution networks. By analysing correlation patterns in smart meter measurements, the service delivers phase assignments, confidence scores, and aggregated phase-load statistics for transformers and feeders.

The AI-Based Network Model Calibration service implements a data-driven calibration process that links real PQM measurement data to a PowerFactory model and iteratively adjusts uncertain profile parameters until deviations between simulation and measurement are minimised. The service accounts for limited observability by applying direct measurement-driven calibration where sensors exist and structured prior knowledge where they do not, preventing overfitting while ensuring physically consistent estimates across the network.

The AI-Enhanced Model Predictive Control service extends the hybrid modelling approach of Service 8 with a forward-looking optimisation component. By forecasting grid states over a short prediction horizon and evaluating different configurations of controllable assets (batteries, curtailable generators, flexible loads) through repeated PowerFactory simulations, the service identifies the asset configuration that best stabilises the grid. All recommendations are presented to the operator for final decision — automated intervention is explicitly not foreseen.

The service processes and analyses asset nameplate images from substation equipment using AI-based computer vision to automatically extract technical parameters. It converts unstructured visual information into structured, machine-readable data, enabling faster and more accurate asset data inventory and supporting the digitalisation of asset management processes.

The service automates the generation of structured analysis reports for disturbance events recorded by protection relays. Using AI, it converts pre-processed and interpreted fault records (from Service 1) into professional expert reports that consolidate all relevant event information. This relieves protection system experts from manual report writing tasks, enabling faster and more consistent analysis of fault and non-fault events.

This service enables processing and analysis of drone-acquired imagery using AI to detect transmission system assets, support inventory identification, and identify potential physical anomalies. By automating visual inspection of power lines and towers, the service reduces the workload of field experts who currently perform inspections manually and enhances asset management databases with detected inventory data.

The Predictive Maintenance service for Energy Efficiency predicts equipment failures and energy efficiency degradation in manufacturing with 7-14 days of advance warning. By detecting conditions that degrade energy performance before causing functional failure, the service enables manufacturers to schedule maintenance during planned production windows, extend equipment lifetime, and maintain equipment in peak energy efficiency condition.

The Manufacturing Energy Demand Forecasting service predicts electrical energy consumption for manufacturing facilities at hourly, daily, and weekly horizons using AI models (Prophet, gradient boosting, LSTM, Transformers). The service enables procurement optimisation, peak demand charge avoidance, renewable energy integration, demand response programme participation, and production scheduling around favourable energy pricing.

The Transformer Thermal and Vibration Anomaly Detection service monitors distribution transformers using vibration measurements, infrared thermal images, and aligned operating data to detect deviations from normal behaviour and identify early signs of potential mechanical or thermal problems. The service supports preventive maintenance planning and provides an additional health-related signal for asset monitoring.

The Power Quality-Based Transformer Health Intelligence service uses power quality and load data from transformer monitoring devices to calculate a Transformer Unit Health Index, detect abnormal trends, and identify early signs of degradation. The service supports more proactive maintenance planning, helps prioritise higher-risk units, and contributes to reducing unplanned outages across the distribution fleet.

The Smart GIS Cable Routing service helps distribution network planners identify and evaluate optimal cable routes using GIS data, infrastructure layers, orthophotography, LiDAR imagery, and terrain information. The service applies optimisation methods to generate feasible cable routes balancing cost, technical constraints, and spatial conditions, delivering editable route proposals directly inside the GIS environment.

The Grid-Aware Flexibility Validation & Compliance Monitoring service provides an auditing layer that evaluates whether flexibility capacity committed by an energy community or aggregator was delivered in accordance with grid constraints, contractual obligations, and operational safety requirements. The service transforms activation records and metering data into structured compliance reports for regulatory and contractual settlement purposes.

The Degradation-Aware Lifecycle Optimisation service integrates battery aging models directly into operational and strategic decision-making for EV batteries and stationary storage. By embedding degradation cost into dispatch decisions, the service enables operators to evaluate energy flexibility strategies not only for short-term economic performance but also for long-term battery health, lifecycle cost, and sustainability — avoiding premature battery replacement from aggressive operation.

The Energy Community Sharing & EV Charging Flexibility Optimisation service implements an AI-driven forecasting and MPC-based optimisation framework to align EV charging demand with locally generated PV electricity and market prices. Using a rolling-horizon Model Predictive Control approach updated every 15 minutes, the service increases renewable self-consumption, reduces peak grid imports, and minimises export losses within energy communities.

The Predictive Maintenance service for EV Charging Points implements a monitoring and prognostics framework that detects abnormal behaviour, emerging faults, and degradation patterns in EV charging infrastructure before they cause service interruption. Combining rule-based diagnostics with data-driven anomaly detection, the service supports operators in reducing charger downtime, improving maintenance planning, extending charger lifetime, and ensuring safer and more reliable charging services.

The School Water Consumption & Anomaly Detection service provides the Municipality of Athens with continuous monitoring of water consumption across instrumented schools, using occupancy-adjusted baselines to detect anomalous flows, background leaks, and wastage events during unoccupied periods. The service makes water losses visible as they develop, giving facility managers the information needed to act before waste becomes financially significant.

The Building Occupancy & Operational Schedule Intelligence service infers actual occupancy patterns from electricity consumption signatures, detects systematic mismatches between declared operating hours and real building use, and generates recommended schedule adjustments to eliminate energy waste during unoccupied periods. By comparing actual consumption behaviour against official schedules, the service identifies buildings where energy continues to be consumed significantly beyond their intended operating periods.

The PV Performance & Self-Consumption Optimisation service addresses the operational management and performance assurance of photovoltaic installations across municipal buildings, while simultaneously maximising the proportion of locally generated solar electricity consumed within each building. The service combines inverter-level fault detection and degradation monitoring via direct API integration with a real-time load alignment capability that shifts flexible building loads into solar generation windows to increase self-consumption.

The Long-Term Energy & Cost Planning service provides the Municipality of Athens with a strategic planning capability that translates operational consumption data into annual and multi-year energy budget forecasts. By constructing building-level consumption trajectories based on actual historical behaviour and seasonal patterns, the service produces portfolio-level annual cost projections under conservative, central, and adverse price scenarios, supporting municipal finance directorates and the central energy management team.

The Short-Term Building Load Forecasting service delivers probabilistic electricity consumption forecasts at 15-minute resolution for horizons spanning from one hour to twenty-four hours ahead, covering every smart-metered building in the municipal estate. Accurate short-term forecasts serve as the informational foundation for demand response scheduling, PV self-consumption optimisation, and anomaly detection baselines across the broader service portfolio. The service maintains a dedicated forecasting model per building, calibrated to each building's specific contextual drivers.

The School Heating Asset Health & Fuel Optimisation service provides continuous thermal comfort compliance monitoring and heating system health assessment for municipal schools in Athens. By exploiting thermostat sensor data combined with outdoor weather observations and electricity consumption telemetry, the service builds a continuous picture of indoor thermal conditions, classifies each monitored interval as compliant or non-compliant, and identifies heating system failures, fuel availability issues, and equipment degradation patterns.

The Energy-Driven Retrofit & Investment Prioritisation service provides the Municipality of Athens with a systematic, data-driven framework for prioritising energy-related capital investments across its public building portfolio. It replaces ad hoc decision-making with a rigorous, continuously updated investment ranking grounded in actual consumption data, peer benchmarking, and waste persistence analysis. The output is a prioritised investment register that identifies which buildings offer the greatest energy saving potential and are best positioned to secure external co-funding.

The Real-Time Energy Anomaly & Waste Intelligence service equips the Municipality of Athens with a continuous, automated intelligence layer for detecting abnormal electricity consumption behaviour and quantifying energy waste across its public building portfolio. Rather than waiting for anomalies to surface through quarterly bill reviews, the service identifies deviations from expected consumption as they emerge in the live meter feed, ranks them by financial impact, and presents building operators with a prioritised action queue. The service builds an evolving waste register that supports both operational decision-making and strategic reporting on energy efficiency progress.

The Real-Time Power Management for TSO-DSO Operations service uses an integrated TSO-DSO network model to analyse how planned or simulated flexibility activations (particularly BESS) influence voltages, power flows, and disturbance propagation across the grid boundary. The service supports better coordination between transmission and distribution operators by helping both sides assess risks earlier and develop more coordinated power management strategies.

The ML-Based Outage Root Cause Identification service uses machine learning to identify patterns in large volumes of SCADA event logs and relate ongoing outage events to previously resolved incidents. By comparing current event sequences with historical incidents, the service returns the most similar past cases with their likely root causes, supporting faster interpretation of outage situations and more consistent disturbance assessment by distribution system operators.

The Dynamic AI-Enhanced Transmission Grid Stability Assessment service delivers data-driven analytical capabilities for assessing the stability of power transmission systems based on instantaneous grid state snapshots. Unlike conventional simulation-based approaches, the service uses similarity-based analytics to retrieve historically or synthetically observed grid states resembling the current operating condition, providing probabilistic and evidence-based insights into potential disturbances and transient instability risks.

This service focuses on the interpretation and advanced analysis of disturbance events recorded by protection relays within the transmission power system. Using AI, the service processes and contextualises fault records to support protection system experts in detecting, reviewing, labelling, and classifying fault and non-fault events. By systematically organising protection relay event data, the service provides continuous situational awareness of transmission system conditions and enables consistent event classification across the network.

The Performance Evaluation service quantifies forecasting performance for offshore renewable energy assets by comparing forecasted and measured production. It identifies systematic errors, detects bias, and provides standardised performance metrics to support continuous improvement of forecasting models. The service supports both real-time monitoring and historical back testing of forecast accuracy.

The Fault Classification and Location service detects, classifies, and localises faults in offshore renewable energy assets using real-time telemetry data. The service identifies abnormal behaviour across electrical, mechanical, and operational signals, assigns fault types to predefined categories (voltage anomalies, current imbalances, converter faults, cable faults, sensor issues), and estimates likely fault locations within the asset or subsystem.

This service classifies and manages the operational states of offshore renewable energy assets — normal, derated, and safeguard — using AI-based classification of real-time telemetry and environmental conditions. When enabled, the service can issue control actions via API to adjust asset operation, supporting safe and efficient management that reduces wear and improves asset lifetime.

The Anomaly Detection for Preventive Maintenance service analyses telemetry data from offshore renewable energy assets to detect deviations from expected operational patterns and estimate the likelihood of component degradation or failure. By providing early warning capabilities, the service enables timely maintenance interventions that reduce downtime and extend asset lifetime.

The Wave Power Generation Short-Term Forecasting service delivers AI-enabled short-term forecasts of electrical power generation for offshore wave energy assets, covering horizons from 0 to 48 hours ahead. Using historical generation data, real-time telemetry, and environmental conditions from the Aguçadoura test site in Portugal, the service supports operational planning, grid balancing, and market participation of offshore renewable energy assets.

The Cascaded Hydropower Forecasting service provides integrated short-term forecasting for the Aliakmonas River hydropower cascade in Northern Greece. Track 1 predicts natural inflows to Ilarionas reservoir; Track 2 predicts electrical production at Asomata HPP, addressing the unique challenge of forecasting in an incomplete cascade where operational data from the intermediate Polyfyto reservoir is unavailable. Both tracks operate at daily resolution over a 1-7 day horizon.

The PV Production Forecasting service delivers short-term and day-ahead photovoltaic power production forecasts for PPC's portfolio of PV generation sites. Using 15-minute multi-sensor site telemetry — including inverter measurements, irradiance, soiling sensors, and solar geometry features — the service provides uncertainty-quantified production estimates at site and inverter level, supporting imbalance management, grid constraint anticipation, and operational planning.

The Wind Farm Event Detection service delivers automated identification, classification, and energy-loss quantification of operational events in wind turbine SCADA streams. Events are categorised into four classes: Environmental, Manufacturer, Utility, and Grid. The service enables systematic comparison of event patterns across turbines and seasons, shortens detection-to-awareness latency, and feeds availability dashboards and energy loss accounting workflows.

The Pumped-Storage Optimisation service provides intelligent operational scheduling for the Sfikia-Asomata coupled hydropower system, optimising bidirectional Sfikia pumped-storage operations to maximise system-wide economic value under varying electricity market prices and water availability conditions. The service operates in two modes: day-ahead scheduling (24h daily) and real-time adjustment via rolling 48h Model Predictive Control updated every 6 hours.

The Wind Power Short-Term Forecasting service delivers probabilistic day-ahead and intra-day wind power forecasts at turbine and farm level using LightGBM ensembles trained on multi-model NWP inputs and SCADA telemetry. The service addresses three interconnected challenges: NWP model selection and weighting, non-linear power curve mapping, and SCADA integration for curtailment and maintenance state accounting. Forecasts are generated in day-ahead (13-40h), extended (40-84h), and intraday rolling (0-12h) modes.

The Sustainable Supply Chain Optimiser generates optimised supply chain plans for multi-stage networks by simultaneously minimising cost, carbon emissions, and lead times while satisfying physical constraints. The service integrates sustainability metrics — including energy consumption and CO2 emissions per supplier and transportation mode — alongside traditional cost and service level objectives, helping organisations meet sustainability targets without compromising operational efficiency.

The AI Manufacturing Process Digital Twin creates virtual representations of physical manufacturing processes using real-time sensor data fusion and continuously updated AI models. The service enables real-time monitoring of process state, simulation of parameter changes, and generation of optimisation recommendations — allowing operators to prevent defects, reduce energy waste proactively, and improve process quality through continuous what-if analysis.

The Energy-Efficient Process Planning Optimiser generates energy-efficient manufacturing process plans that optimise the selection of manufacturing processes, their sequences, machine assignments, and process parameter values. By simultaneously minimising energy consumption, cycle time, and manufacturing cost while ensuring quality compliance, the service embeds energy efficiency into the planning phase of production before manufacturing begins.

The AI-Assisted Production Scheduling service generates optimised production schedules that allocate and sequence manufacturing tasks across available resources, with energy efficiency as a primary optimisation criterion alongside makespan and due-date compliance. By evaluating alternative machine sequences and task assignments using AI and combinatorial optimisation, the service identifies schedules that minimise total energy consumption for equivalent production output.

The AI-Driven Energy Management service optimises power allocation between fuel cells and batteries in Fuel Cell Hybrid Electric Vehicles (FCHEVs) using a reinforcement learning framework. The service minimises hydrogen consumption, maintains battery SoC, and extends powertrain component lifetime through adaptive power-split decisions evaluated over complete driving cycles.

The AI-Driven Active Control service provides an adaptive control framework for hydrogen technologies within renewable energy microgrids, optimising the coordination between electrolysers, fuel cells, and energy storage systems to ensure efficient energy flow between renewable generation, hydrogen production, storage, and consumption. The service aims to maximise renewable utilisation and improve microgrid flexibility and resilience.

The Data-Driven Predictive Modelling service provides an AI-based framework for predicting the performance evolution and degradation of hydrogen fuel cell systems using time-series operational data. Based on LSTM networks with self-attention mechanisms, the service supports condition monitoring, predictive maintenance, and lifecycle optimisation by delivering reliable forecasts of stack voltage as the primary performance health indicator.

The AI-Enhanced Multi-Agent Testing for V2G Applications service provides a simulation and validation environment for evaluating decentralised V2G control strategies based on AI-driven multi-agent systems. The service models EV fleets interacting with energy markets, grid signals, and operational constraints, enabling pre-deployment validation of decentralised coordination strategies including stability analysis, fairness assessment, and robustness testing.

The Battery & EV EMS Optimisation service provides coordinated optimisation of stationary battery systems and EV charging assets to maximise PV self-consumption within energy communities. Using a rolling-horizon MPC framework updated every 15 minutes, the service generates coordinated operational schedules for battery dispatch and EV charging that respect technical constraints (battery SoC bounds, EV departure requirements, charger power limits) while maximising renewable energy utilisation.

The Wind Generation Forecasting service provides high-resolution short-term and day-ahead wind power production forecasts for EV-integrated energy communities, supporting battery scheduling and — when EV telemetry becomes available — EV-aware flexibility coordination. Currently deployed for the Beckerich energy community with a 4.2 MW wind asset and 60 kW/160 kWh battery, the service is designed to be battery-ready and EV-ready.

The Localised PV Generation Forecasting service provides high-resolution photovoltaic generation forecasting for energy communities, combining data-driven machine learning with physics-informed feature engineering. The service produces deterministic and probabilistic forecasts at 15-minute resolution for short-term and day-ahead horizons, enabling community operators to anticipate renewable availability and improve scheduling of flexible loads.

The EV-Driven Demand Forecasting service predicts both aggregate and disaggregated electricity demand in residential energy communities with significant EV adoption, explicitly accounting for EV charging behaviour, household consumption patterns, and external contextual factors. The service supports operational planning, demand response strategy design, and self-consumption optimisation for DSOs, aggregators, and community managers.

The EV-User Charging and Usage Profiles Prediction service estimates when EV users are likely to charge, where charging activity is expected to occur, and how much energy is likely to be consumed during future sessions. By incorporating behavioural, spatial, and contextual factors, the service provides actionable forecasts that support infrastructure planning, congestion mitigation, smart charging strategy design, and flexible demand coordination.

The Digital Twin for DHN Optimisation provides a virtual representation of the Torrelago district heating network, enabling advanced monitoring, simulation, and scenario-based optimisation of thermal energy distribution. By integrating real-time and historical data with demand forecasting outputs, the digital twin allows evaluation of alternative operational strategies (supply temperature changes, load distribution, control setpoints) without impacting real system operation.

The Heating Demand Forecasting service uses AI-based models (LSTM and XGBoost) to forecast short-term thermal energy demand in the Torrelago district heating network. By integrating historical consumption patterns with external temperature forecast datasets, the service enables proactive decision-making and improved planning of heat production and distribution across the network.

The Heating Consumption Optimisation service analyses thermal energy consumption within the Torrelago district heating network to identify inefficiencies, detect anomalous consumption patterns (including elevated return temperatures and abnormal demand during mild conditions), and optimise heat distribution across connected buildings. Using statistical analysis, contextual peak detection, and AI models (LSTM, XGBoost), the service provides actionable insights for energy efficiency improvements.

The Building Flexibility & Demand Response Optimisation service uses AI to actively manage electricity consumption across a municipal building portfolio in response to external grid requests, dynamic tariffs, and cost signals. By analysing incoming consumption data alongside grid operator notifications and tariff conditions, it generates actionable curtailment and rescheduling recommendations that building operators can review and approve before implementation. The result is a systematic approach to demand response participation that reduces peak-hour expenditure and positions the municipal estate as a controllable flexibility asset.

The AI-State Estimation service delivers high-resolution near-real-time representations of the operational state of LV/MV distribution grids. By combining AI-generated load profiles for unmetered assets with measured feeder-level SCADA data and a full power-flow solver, the service computes voltage magnitudes, angles, active/reactive power flows, and currents for every node and branch including those without direct measurement.

The AI-based State Estimation Forecast service delivers 0-48 hour predictions of the full electrical state of a distribution grid, combining AI load/generation forecasting with a physical power-flow solver to produce future values of active/reactive power flows, voltage magnitudes, and asset utilisation for every network component. The service enables DSOs to anticipate congestion, thermal overloads, or voltage violations before they occur.

The AI-Enhanced Model-Based Ancillary Services service implements a hybrid approach combining physics-based simulation (PowerFactory) with data-driven analysis to identify systematic deviations between the network model and real-world measurements. By learning recurring deviation patterns — at specific times of day, load profiles, or seasonal conditions — the service applies a correction layer on top of the physical model to improve reliability of model-based grid management processes.