De-noising of Vibration Signals of Induction Motor using Compressed Sensing.

EasyChair Preprint 2081

Abstract

This paper proposes a de-noising method for vibration signal based on compressed sensing. Compressed sensing technology is a method for de-noising a vibration signal, in the field of signal processing. It can achieve data acquisition as well as data compression at the same time. In this paper first, the original signal express into a low dimensional space. According to the compressed sensing theory the uncontaminated vibration signal can be represented sparsely by some transform domain while noise can’t be represent, therefore the noise information in original signal can be banished by compressed sensing. Then the original signal can be recovered by reconstruction algorithm. The orthogonal matching pursuit algorithm used in this paper and finally de-noising is achieved. This method is verified by vibration signal of induction motor which polluted by white Gaussian noise. The simulation results show that the SNR could be improved and signal reconstruction error is minimum when we set appropriate value of sparsity and linear measurement in OMP algorithm.

Keyphrases: Compressed signal, De-noising, Gaussian Random Matrix, OMP algorithm, dictionary matrix, sparse signal, vibration signal

@booklet{EasyChair:2081,
  author    = {Niraj Kumar},
  title     = {De-noising of Vibration Signals of Induction Motor using Compressed Sensing.},
  howpublished = {EasyChair Preprint 2081},
  year      = {EasyChair, 2019}}