The practicability of ICA in home appliances load profile separation using current signature: A preliminary study

The identification of appliances has its unique application in detecting fraudulent electricity consumption; this is an important feature for smart meter. The current signature of signal individual and composite appliances has a possible feature set to separate an individual load from the current signature of composite load. We exploited the source signal separation feature of Independent Component Analysis (ICA) to observe the feasibility of ICA in separating an individual load profile from its composite current waveform. The results suggest that, for most of the electrical loads(used for experiment) ICA has been successful to extract the individual waveform from its composite load current signature with more than cross correlation value of 0.75, between the experimental current signature and ICA extracted current signature. The research work proposes a direction to use ICA as an intermediate step to develop a device identification classifier.