Services Catalogue

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.