A Novel Flexible and Scalable Nonintrusive Load Monitoring Approach Using Wavelet Design and Machine Learning

This paper addresses the problem of non-intrusive appliance load monitoring considering the flexibility of adapting to a new load and the scalability when adopting more appliances. The proposed approach in this paper uses the concept of wavelet design to generate new wavelets to match the features in the transient signals of such loads. The detection and the classification processes are then automated through the application of the unsupervised clustering and the decision tree classifier. The proposed concept has been experimentally verified using a real-test system. The flexibility and the scalability of the proposed method have been discussed and have been exemplified through numerical results.