Residential load signature analysis for their segregation using wavelet—SVM

The unique power consumption pattern of each appliance or a combination of appliances can be analyzed using their load signatures which can be acquired from a single point. It is quite difficult to disaggregate the similar kind of home appliances because of their similar characteristics. Wavelet coefficients of load signature have been chosen as the feature vectors which reflected the edge over other features. These coefficients serve as input data for the classifier. By considering various classification algorithms a comparison has been made and the best algorithm was investigated which is the linear Support Vector Machine (SVM) for the selected similar appliances. The results of laboratory experiment promise a new application for smart meters.