A Method Based on Independent Component Analysis for Adaptive Decomposition of Multiple Power Quality Disturbances

摘要

This paper proposes a method based on Independent Component Analysis (ICA) to adaptively decompose signals containing multiple power quality disturbances. The method aims to decompose the power system signal (voltage or current) into its isolated disturbances when a multiple disturbance situation occurs, making it possible to obtain more specific information about the different disturbances that can be occurring simultaneously. ICA is originally a multichannel technique. However, the method proposes to use it to blindly separate the different disturbances existing in a single measured signal (single channel). For this purpose, a fixed filter bank is used as a pre-processing step. We demonstrate that the overall response of the proposed method corresponds to an adaptive linear filter bank, and show that as long as the multiple independent disturbances are spectrally disjoint from each other, the method can achieve satisfactory results in their separation. The proposed method is evaluated by means of its application to synthetic, as well as to actual data. A comparison with a method based on the discrete wavelet transform for the same aim shows that the proposed method achieves better results...