Many properties have unnecessarily high energy consumption. Additionally, there are no effective tools to balance consumption across a large portfolio of properties, resulting in a maximum power demand that is higher than it could be with efficient tools. Our aim is to contribute to reduced energy consumption in large property portfolios by increasing our understanding of how energy is used.

Artificial Intelligence (AI) and Machine Learning can be used to identify when a building changes its energy usage patterns, both individually and in comparison to similar buildings. This involves forecasting energy usage through time series analysis of historical and current data and identifying root causes for deviations.

Various datasets will be included and analyzed using machine learning algorithms and/or methods to create energy profiles for properties. Data include historical energy usage, construction data about the property, operational data (e.g., how and when the building is used), weather data, and real-time data from building automation systems.

Based on the increased understanding the analyses will offer, we will be able to estimate future energy usage and identify deviations, such as when a property uses more energy than before, for example due to faults in the ventilation system. We will also be able to detect differences in energy usage between similar properties.

A system to visualize energy usage will be developed. Data will be sourced from properties in the municipalities of Skellefteå and Piteå, and from AB Kristianstadsbyggen, as well as energy companies Piteå Energi, Skellefteå Kraft, and C4 Energi. We will evaluate the results in the municipalities of Skellefteå, Piteå, and Kristianstad for the properties addressed in the project.