Appliance level stand-by burst forecast modelling using machine learning techniques
Electric power is an expensive and scarce resource and the concept of modern life is not possible without the continuous uninterrupted supply of it. Therefore, a lot of efforts have been made in past to conserve and optimize the use of electric power so that it could be efficiently distributed to al...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2020
|
| Subjects: | |
| Online Access: | http://psasir.upm.edu.my/id/eprint/104212/1/ABID%20MUSTAFA%20-%20IR.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Electric power is an expensive and scarce resource and the concept of modern life is not possible without the continuous uninterrupted supply of it. Therefore, a lot of efforts have been made in past to conserve and optimize the use of electric power so that it could be efficiently distributed to all consumers. The efforts to conserve the energy include government and other organizations' sponsored awareness campaign for public to encourage them to use the best practices while the efforts for optimizing its use are led by the researchers and industries. The electrical appliances and equipment are developed in a way that optimize the use of energy. In this direction, one of the important inventions was the use of standby mode for the electrical appliances which is employed when the appliance is plugged-in but not in active use. The standby mode helps optimize electric power use yet it causes some power leakage. This study strives to forecast the appliances' state (standby or running) in next minutes to prevent the power leakage during the standby mode: by accurately forecasting the standby burst the appliance could be put in off state during the forecasted burst duration. This work proposes a technique to model power consumption data and presents a comparative study of five different machine learning algorithms to study their suitability to forecast an appliance's state and standby burst. The proposed approach achieved around 90 percent accuracy and very good indications over precision, recall and F1-Score for models built using Decision Tree, Logistic Regression, Support Vector Machine (SVM), K- Nearest Neighbor (KNN), and Multilayer Perceptron (MLP). |
|---|