Predicting HVAC Energy Consumption

HVAC systems are large consumers of energy, resulting in high electricity bills for commercial and office buildings. Hence, HVAC systems are expected to play a central role in demand-response programmes for office buildings. Estimating the power consumption of HVAC systems is therefore of paramount importance to assess the potential savings that can be achieved by operating them more efficiently.

The goal of this semester project is to build a data-driven model to estimate HVAC consumption in office buildings. Two different approaches will be investigated. In the first one, a model is built offline (using historical data) and then deployed for evaluation. In the second, more sophisticated, approach, an initial model is built offline and continuously adapted by taking into consideration new data. Whenever possible, we will not implement machine learning algorithm from scratch. Instead, we will use public machine learning libraries available.

If you have any question, just drop us an email, or come to our office:

• Matteo Vasirani (BC114): [email protected]
• Tri Kurniawan Wijaya (BC147): [email protected]

Prerequisites

• Having the spirit of data exploration
• Familiar with (or at least having desire to learn) data mining/machine learning techniques

Possible starting date: asap.