Research on power quality signals reconstruction method based on K-SVD dictionary learning

With the rapid development of modern science and continuous innovation of technology, a large number of power quality data storage and transmission has caused great burden to power system operation, so it is of great practical significance to study power quality. In order to solve the problem of the existing fixed orthogonal sparse basis is not enough to represent the unknown power quality signals flexibly when using the compression sensing theory to reconstruct power quality signals, which leads to poor signal reconstruction effect and poor applicability. K-SVD dictionary learning is introduced into the power quality signals reconstruction. Firstly, a large number of power quality signals samples are trained by K-SVD dictionary to get a super complete dictionary. Secondly, Gauss random matrix is selected as the measurement matrix. Lastly, KSVD-CoSaMP algorithm is proposed to reconstruct the power quality signals. Simulation experimental results show that, power quality signals are reconstructed using the proposed algorithm compared DCT algorithm, the SNR of all the reconstructed signals are higher than 30 dB, ERP are over 99.6%, and MSE are under 3%. The indexes of the proposed algorithm are obviously superior to the existing DCT algorithm, which further verifies the superiority and applicability of the proposed algorithm.