Energy spectrum-based wavelet transform for non-intrusive demand monitoring and load identification

Though the wavelet transform coefficients (WTCs) contain plenty of information needed for turn-on/off transient signal identification of load events, adopting the WTCs directly has the drawbacks of taking a longer time and much memory for the identification process of non-intrusive load monitoring (NILM). To effectively reduce the number of features representing load turn-on/off transient signals, an energy spectrum of the WTCs in different scales calculated by the Parseval´s Theorem are proposed and presented in this paper. The back-propagation (BP) classification system is then used for artificial neural network (ANN) constructions and load identifications. The high successful rates of load events recognition from both experiments and simulations have proved that the proposed algorithm is applicable in multiple load operations of non-intrusive demand monitoring applications.