Non-intrusive television recognition based on harmonic current characteristics

Non-intrusive load monitoring (NILM) is an effective means to realize detailed user energy monitoring. As a kind of appliances which is widely used and can intuitively reflect the user's behavior, the recognition of television (TV) is of great significance for NILM application. However, it is often difficult to recognize TV in the aggregated load data because of its relatively poor power-range distinction and complex power consumption characteristics. In this paper, the harmonic current characteristics or the fundamental-third harmonic current space trajectory is analyzed, and they are combined with the random power fluctuation characteristics during TV operation and other prior knowledge as multi-dimensional load signatures of TV. Based on this, an unsupervised non-intrusive TV recognition framework and method is proposed. Experimental analysis on collected real-world dataset shows that the proposed method can detect and recognize TV with fluctuating operation characteristics accurately and effectively.