Electrical Appliance Identification Through Signal Processing of Electrical Wave Signals
Software as a Service
NoCategory
Thesis DesignClients
Techstack
Purpose
A Mapua University thesis to build a non-intrusive appliance identification system: acquire electrical signal data through a working prototype, train a Convolutional Neural Network on it, and test whether the system correctly identifies five appliance types.
Description
An Arduino-based prototype samples current draw from connected appliances, and a Python pipeline converts the captured waveforms into signal images for classification. A CNN built on TensorFlow — including a MobileNetV2 transfer-learning variant — is trained on the gathered dataset, with dedicated scripts for data gathering, training, prediction, and unauthorized-device checks. Model performance is documented through accuracy/loss curves and confusion matrices produced during evaluation.
