Edition Aug 4th-5th, 2023

NOI Hackathon Summer Edition



With this project, we aim to give access to energy consumption management.

We trained a ML model that predicts energy consumption and production. 

On one hand it was trained on a dataset composed of two-year monitoring of energy consumption and production of an office building in central Italy (Terni). The building has HVAC system (Heating, Ventilation, Air Conditioning), heat pumps for space heating /cooling (overall 120-140 KW load) and lighting subsystems controlled individually and/or overall by BMS. It is also part of a small smart-grid which includes a PV (Photovoltaic) plant (180KW). This dataset underwent a cleaning and normalisation process.

On the other hand it was trained on passed data of weather conditions, specifically temperature, solar inclination and solar diffused and solar direct radiation (which translate into solar irradiance). 

The model makes the predictions also based on weather forecast with the same measures.

The model is connected through a pipeline to a dashboard that can be accessed by the citizen. Energy consumption and production is suitably visualised for both the past years (back to August 2018) and as a prediction for the following day.