Thermal Profiling of Residential Energy Consumption for Heat Pump and District Heating Customers
|Title||Thermal Profiling of Residential Energy Consumption for Heat Pump and District Heating Customers|
|Summary||Finding interesting patterns of heat pump and district heating customers in order to identify meaningful thermal profiles|
|Keywords||interesting pattern discovery, energy optimization, data mining, machine learning|
|References|| H. Gadd and S. Werner, "Heat load patterns in district heating substations", Applied Energy, vol. 108, pp. 176-183, 2013.
A. Albert and R. Rajagopal, "Thermal Profiling of Residential Energy Use", IEEE Transactions on Power Systems, vol. 30, no. 2, pp. 602-611, 2015.
A. Albert and R. Rajagopal, "Building dynamic thermal profiles of energy consumption for individuals and neighborhoods", 2013 IEEE International Conference on Big Data, 2013.
|Prerequisites||Artificial Intelligence and Learning Systems courses; good knowledge of data mining; programming skills for implementing machine learning algorithms|
|Supervisor||Ece Calikus, Sławomir Nowaczyk|
Nowadays large volumes of energy data are continuously collected through a variety of smart meters from different environments. Such data have a great potential to influence the overall energy balance of our communities by identifying thermal behaviour and optimizing building energy consumption and by enhancing people’s awareness of energy wasting. Modelling thermal behaviours of households, buildings and substations are key elements to estimate heat demand and identify normal-abnormal energy consumption. Gadd and Werner (2013) have conducted such study by manually analyzing district heating customers in different categories. They identified four different profiles as a result:
-Continuous operation control by profiling continuous activity of different type of buildings
-Night setback control by profiling night by profiling when the set point for the indoor temperature is lowered during the night
-Time clock operation control 5 days a week by profiling daytime and weekdays use of buildings
-Time clock operation control 7 days a week by profiling daytime and 7 days a week use of buildings.
In addition to those, modelling thermal behaviours can capture things like, for example, presence at home, holiday usages, seasonal sensitivity to the outside temperature, continuous operation, night set back operation etc. They are key elements to estimate heat demand and identify normal-abnormal energy consumption.
However, it is extremely costly to extract such patterns from various customers by manually. In this project, we aim to automatically find interesting behavioural patterns of heat pump and district heating customers in order to identify meaningful thermal occupancy or building profiles by applying data mining and machine learning approaches.
- Data exploration: Analyzing and visualizing data collected from smart meter readings of district heating and heat pump customers
- Thermal profiling: Modelling different thermal profiles based on spatial behaviours (Helsinborg customers, Halmstad customers etc.), temporal behaviours (holidays, seasons etc. ), building categories, billing amounts, etc.
- Evaluation: Finding evaluation strategy to measure interestingness of those profiles. Estimating usefulness of those profiles in other machine learning tasks like anomaly detection or energy demand forecasting of the customers.
We have collaboration with 3 companies in energy domain within this project i.e. HEM, Öresundskraft and EasyServ which give an opportunity to work on solving real-world problems.