Non-intrusive electrical energy monitoring based on intelligent system
The demands of electricity are increasing from day to day as many housing and building construction are built to fulfil the demands. Abreast with the development, the electricity is an essential for daily basis, thus the wastage is also a problem that needs to overcome. The electrical wastage proble...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2015
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/79007/1/ZulhilmanHassanMFKE2015.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The demands of electricity are increasing from day to day as many housing and building construction are built to fulfil the demands. Abreast with the development, the electricity is an essential for daily basis, thus the wastage is also a problem that needs to overcome. The electrical wastage problem is focused on the residency college as students are always not aware to turn off the appliances when leaving the room, for instance, the fan and lamp. The first stage to overcome the wastage problem, an approach called “Non-Intrusive Electrical Energy Monitoring (NIEM)” is proposed to this project. NIEM encompass a method of detecting the electrical energy consumption in a building by using a single set of sensor on the main distribution board for each building. This method is in contrast to Intrusive Electrical Energy Monitoring (IEM) where the end-use devices are sensed. To realize the method used, an energy meter is used to measure the electrical consumption by the appliances. The data obtained will be analyzed using a method called Multilayer Perceptron (MLP) technique of Artificial Neural Network (ANN). The technique will firstly implement the event detection to identify the type of loads and the power consumption of the load which is intensified as fan and lamp. The switching ON and OFF events of the loads are made in order and random to test the capability of MLP to classify the type of loads. Then the data were divided to 70% for training, 15% for testing and 15% for validation. The output of the MLP is either ‘1’ for fan or ‘0’ for lamp. The system can be re-train to obtain a good performance, lower Mean Square Error (difference between output and target), and lower percent error (misclassified data). For later stages in future, a Neural Network system can be design to automatically turn off the appliances whenever not in used, so that the electrical wastage and monthly bill can be reduced to strive for a green and energy saving manner. |
|---|