Non-Intrusive Load Monitoring With Focus On Appliance Electric Signature Using Convolutional Neural Network Enhanced By Generative Adversarial Network

Nuha, Muhammad Fajrul Alam Ulin (2024) Non-Intrusive Load Monitoring With Focus On Appliance Electric Signature Using Convolutional Neural Network Enhanced By Generative Adversarial Network. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6026231024-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (2MB) | Request a copy

Abstract

Non-Intrusive Load Monitoring (NILM) is a technology that utilizes machine learning and data analysis to disaggregate and identify individual appliance-level energy consumption patterns from a building’s overall electricity usage. Unfortunately, obtaining datasets for the development of NILM is a significant challenge. Publicly accessible NILM datasets often consists of fewer than 10 households, which is due to the expensive nature of obtaining these datasets. These limited datasets, which are used as training data, lead to the creation of overfitting NILM models. The purpose of this study is to mitigate the limitations associated with insufficient training data for NILM model development. To achieve this, the methodology employs Generative Adversarial Networks (GANs) in the pre-processing phase. Subsequently, a model Convolutional Neural Network (CNN) model is employed to classify the appliances and predict each appliance’s
load. Three CNN classification models were developed and their results were analysed. Model 1 was trained exclusively on actual data, Model 2 on a combination of actual and synthetic data, and Model 3 on the same combination but with
additional overfitting prevention techniques. Our findings show that utilizing GANs significantly improves model performance. The MAE values for Models 1, 2, and 3 were 21.52, 19.61, and 3.02, respectively, while the RMSE values were 158.64, 146.83, and 31.71. The effectiveness of this approach, combined with the absence of additional costs, highlights its potential applications in various domains such as Demand Response and Home Energy Management Systems.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Convolutional Neural Network , Generative Adversarial Network, Non-Intrusive Load Monitoring
Subjects:	T Technology > T Technology (General) > T174 Technological forecasting T Technology > T Technology (General) > T57.5 Data Processing T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK1007 Electric power systems control
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 59101-(S2) Master Thesis
Depositing User:	Muhammad Fajrul Alam Ulin Nuha
Date Deposited:	31 Jul 2024 07:27
Last Modified:	31 Jul 2024 07:27
URI:	http://repository.its.ac.id/id/eprint/109165

Actions (login required)

View Item