Identification of electrical appliances via analysis of power consumption

Identification of small electrical appliances via power consumption requires accurate detection and evaluation of steady-state sections and transient sections. However, the steady-state sections and transient sections are extracted from low frequency sampled (1 Hz) power measurements. We gain the steady-state sections and transient sections by processing real and reactive power measurements with a robust bucketing technique and unsupervised clustering. Macroscopic features for detected steady-state sections and transient sections are then extracted. Besides, our method estimates the similarity of steady-state sections and transient sections and determines recurred sections accurately. The proposed method is applicable for an inexpensive, unsupervised learning of small electrical appliances in real time.