Abstract
In a building, electrical energy is the main source of energy used, therefore proper control and management of it provide benefits to both the owners and the building occupants. Abnormal behavior in the electrical consumption of the building can generate supply problems or, on the contrary, oversizing of the energy supply, which in turn generates problems in the medium and long term, so detecting these abnormal behaviors is of vital importance. Currently, the operation, planning, and energy management in buildings have taken great importance to optimize resources in them, for this, energy efficiency policies and systems are implemented. Recent research shows that electricity consumption patterns allow observing energy demand behaviors in a facility and, in turn, implementing strategies to increase energy efficiency, Therefore, evaluating the changes in the consumption profiles allows associating these profiles with possible abnormal consumption events in a facility. This can be used to generate alarms, reduce maintenance costs, and respond quickly to the presence of some type of abnormal consumption. For this reason, many efforts have been made to define techniques and methods to evaluate changes in energy consumption. However, the time series components of the electricity demand of each building are a considerable factor in the energy analysis, which in turn prevents the creation of a general method or technique to detect abnormalities. Therefore, this research proposes the use of an intelligent agent trained through reinforcement learning to detect abnormalities in the monitoring of electricity consumption in a building, tracking the components of time series of electricity typical of said building. This would significantly improve the ability to detect anomalies in the electricity consumption profiles since the agent will adapt to the temporary conditions of each building. This method proposes to present a multi-criteria interpretation that explains the possible cause of abnormal consumption. This real-time technique can help reduce costs and energy consumption, as well as quickly detect abnormal consumptions and faults, as well as help, improve energy consumption and help establish whether energy-saving policies in a building are effective.
Original language | English |
---|---|
Title of host publication | Advances in Artificial Intelligence, Software and Systems Engineering - Proceedings of the AHFE 2021 Virtual Conferences on Human Factors in Software and Systems Engineering, Artificial Intelligence and Social Computing, and Energy, 2021 |
Editors | Tareq Z. Ahram, Waldemar Karwowski, Jay Kalra |
Publisher | Springer Science and Business Media Deutschland GmbH |
Pages | 207-215 |
Number of pages | 9 |
ISBN (Print) | 9783030806231 |
DOIs | |
State | Published - 2021 |
Event | AHFE Conferences on Human Factors in Software and Systems Engineering, Artificial Intelligence and Social Computing, and Energy, 2021 - Virtual, Online Duration: 25 Jul 2021 → 29 Jul 2021 |
Publication series
Name | Lecture Notes in Networks and Systems |
---|---|
Volume | 271 |
ISSN (Print) | 2367-3370 |
ISSN (Electronic) | 2367-3389 |
Conference
Conference | AHFE Conferences on Human Factors in Software and Systems Engineering, Artificial Intelligence and Social Computing, and Energy, 2021 |
---|---|
City | Virtual, Online |
Period | 25/07/21 → 29/07/21 |
Bibliographical note
Publisher Copyright:© 2021, The Author(s), under exclusive license to Springer Nature Switzerland AG.
Keywords
- Changes in consumption
- Electrical consumption patterns
- Electrical consumption profiles (ECP)
- Outlier detection
- Reinforcement Learning