Smart Building Energy Management: Load Profile Prediction using Machine Learning

Smart buildings are gaining popularity with the surfacing trend of smart grid and smart city. Effective energy management is a major aspect of the smart building management system that demands accurate prediction of building electrical energy consumption profile. The paper focuses on a data-driven a...

Full description

Saved in:
Bibliographic Details
Main Authors: Revati, G., Hozefa, J., Shadab, S., Sheikh, A., Wagh, S. R., Singh, N. M.
Format: Conference Proceeding
Language:English
Subjects:
Artificial neural networks
Dynamic Mode Decomposition
Energy Management
Gaussian Process Regression
Predictive models
Process control
Schedules
Smart Building
Smart buildings
Smart cities
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Smart buildings are gaining popularity with the surfacing trend of smart grid and smart city. Effective energy management is a major aspect of the smart building management system that demands accurate prediction of building electrical energy consumption profile. The paper focuses on a data-driven approach to load profile prediction with the highlighted benefit of a model-free environment. The electricity consumption profile of a commercial smart building is predicted using Gaussian Process Regression (GPR), and a comparative study is carried out to highlight the issues associated with Polynomial Regression, Artificial Neural Network (ANN), Dynamic Mode Decomposition (DMD), and Hankeled DMD (HDMD). For testing the effectiveness of the proposed methodology, various test scenarios were conducted and from the result, it is observed that the HDMD and GPR are preferred techniques to provide reliable prediction, which is beneficial for arranging a specific demand response schedule to earn benefits like financial rewards and carbon footprint curtailment.
ISSN:2473-3504
DOI:10.1109/MED51440.2021.9480170