D2.2 Advanced Forecasting Tools

This deliverable presents the advanced forecasting tools developed within the FlexBIT project, focusing on ultra-short-term and short-term prediction of photovoltaic (PV) generation and electricity demand. The forecasting module constitutes a key component of the FlexBIT digital platform, providing predictive capabilities that support real-time control, optimization, and flexibility management across residential, tertiary, and industrial energy systems.

The scope of this deliverable covers data originating from multiple sources, including platform operational data, IoT and SCADA/EMACS measurements, as well as external data streams such as weather services and market signals. The document describes the data sources, feature engineering techniques, forecasting methodologies, model selection strategies, and validation procedures applied across different demonstrators within the project.

The forecasting tools are designed to operate under the specific requirements of the FlexBIT platform, where high temporal resolution, low latency, and robustness to variability are essential. In particular, the models support rolling predictions at minute-level granularity, enabling their integration into the real-time advisory and control loop of the platform.

This deliverable builds upon the system architecture and control concepts defined in Deliverable D2.1, providing the predictive layer that feeds optimization and decision-making modules. It is closely linked with other activities within Work Package 2, including data integration, flexibility identification, and the development of control algorithms.

By combining advanced machine learning techniques with scalable data processing pipelines, the forecasting module contributes to the overall objectives of FlexBIT, enabling improved utilization of renewable energy sources, enhanced