Machine learning techniques for prediction of time series data

This report explores machine learning techniques for electrical load forecasting, which is a critical aspect in the management of modern-day power systems. With the integration of renewable energy sources and the rise of smart grids, the need for accurate load forecasting has become more importan...

Full description

Saved in:
Bibliographic Details
Main Author: Hariharan, Dhruv
Other Authors: Yeo Chai Kiat
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2024
Subjects:
Online Access:https://hdl.handle.net/10356/175510
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
id sg-ntu-dr.10356-175510
record_format dspace
spelling sg-ntu-dr.10356-1755102024-04-26T15:45:28Z Machine learning techniques for prediction of time series data Hariharan, Dhruv Yeo Chai Kiat School of Computer Science and Engineering ASCKYEO@ntu.edu.sg Computer and Information Science This report explores machine learning techniques for electrical load forecasting, which is a critical aspect in the management of modern-day power systems. With the integration of renewable energy sources and the rise of smart grids, the need for accurate load forecasting has become more important. Traditional statistical methods often fall short of capturing the complex, non-linear patterns of electrical load data. While Machine Learning and Deep Learning methods have produced good results, they also come with their own inherent lim itations. To bridge this gap, we explore the application of Transformer models for electrical load forecasting. This research investigates the performance of the Transformer-SWT model across various datasets, aiming to leverage the model’s ability to handle sequential data and the SWT’s proficiency in decomposing time series into components for better analysis. Through extensive experiments involving hyperparameter tuning and comparative analysis against state-of-the-art models, we demonstrate the Transformer-SWT model’s robustness and versatility. The findings suggest that with optimal hyperparameter settings and suffi cient training, the Transformer-SWT model holds the potential to not only match but also surpass current benchmarks in electrical load forecasting. This study contributes to the advancement of machine learning techniques for electrical load forecasting, highlighting the Transformer model’s promising capabilities in forecasting tasks and setting the stage for further research in this field. Bachelor's degree 2024-04-25T07:10:02Z 2024-04-25T07:10:02Z 2024 Final Year Project (FYP) Hariharan, D. (2024). Machine learning techniques for prediction of time series data. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/175510 https://hdl.handle.net/10356/175510 en application/pdf Nanyang Technological University
institution Nanyang Technological University
building NTU Library
continent Asia
country Singapore
Singapore
content_provider NTU Library
collection DR-NTU
language English
topic Computer and Information Science
spellingShingle Computer and Information Science
Hariharan, Dhruv
Machine learning techniques for prediction of time series data
description This report explores machine learning techniques for electrical load forecasting, which is a critical aspect in the management of modern-day power systems. With the integration of renewable energy sources and the rise of smart grids, the need for accurate load forecasting has become more important. Traditional statistical methods often fall short of capturing the complex, non-linear patterns of electrical load data. While Machine Learning and Deep Learning methods have produced good results, they also come with their own inherent lim itations. To bridge this gap, we explore the application of Transformer models for electrical load forecasting. This research investigates the performance of the Transformer-SWT model across various datasets, aiming to leverage the model’s ability to handle sequential data and the SWT’s proficiency in decomposing time series into components for better analysis. Through extensive experiments involving hyperparameter tuning and comparative analysis against state-of-the-art models, we demonstrate the Transformer-SWT model’s robustness and versatility. The findings suggest that with optimal hyperparameter settings and suffi cient training, the Transformer-SWT model holds the potential to not only match but also surpass current benchmarks in electrical load forecasting. This study contributes to the advancement of machine learning techniques for electrical load forecasting, highlighting the Transformer model’s promising capabilities in forecasting tasks and setting the stage for further research in this field.
author2 Yeo Chai Kiat
author_facet Yeo Chai Kiat
Hariharan, Dhruv
format Final Year Project
author Hariharan, Dhruv
author_sort Hariharan, Dhruv
title Machine learning techniques for prediction of time series data
title_short Machine learning techniques for prediction of time series data
title_full Machine learning techniques for prediction of time series data
title_fullStr Machine learning techniques for prediction of time series data
title_full_unstemmed Machine learning techniques for prediction of time series data
title_sort machine learning techniques for prediction of time series data
publisher Nanyang Technological University
publishDate 2024
url https://hdl.handle.net/10356/175510
_version_ 1806059840037978112