Semi-blind Source Separation Based on the Inner Product of the Feature Ranking Vector

To solve the problem of single-channel blind source separation for the aggregated current signal of the multiple home appliances, a novel method of semi-blind source separation extracting the principal component based on inner product of the feature ranking vectors of multiple hidden states was proposed in this paper. The fuzzy feature ranking vector is acquired by setting the corresponding spectrum value to 1 or 0 according to the sequence of the harmonics of the signal, the inner product between the feature ranking vector of observed signal and the feature ranking vector of each standard state is calculated, and the principal components of source signal are extracted according to the maximum inner product step by step. We chose an air conditioner, a microwave oven and a washing machine as the test object to testify the proposed method. The experimental results showed that the aggregated current signal of the multiple home appliances can be fast and accurately separated using this method and which can be widely used in the field of remote Prognostics and Health Management for home appliances.