A noise-reduction neural network as a preprocessing stage in the SVD based method of harmonic retrieval

Author

Rao, Sathyanarayan S. ; Pisharam, Pradeep M.

Author_Institution

Dept. of Electr. Eng., Villanova Univ., PA, USA

fYear

1990

fDate

1-3 May 1990

Firstpage

491

Abstract

A noise-reduction neural network is proposed as a preprocessing stage in the singular value decomposition (SVD)-based state-space approach to harmonic retrieval. By performing noise-reduction on the data set, the performance is improved of the various criteria available in the literature for identifying the significant singular values in the SVD of the estimated covariance matrix. Computer simulations performed using sampled sinusoids in white noise show that the network is capable of learning to perform noise reduction

Keywords

computerised signal processing; harmonics; interference suppression; learning systems; neural nets; state-space methods; white noise; SVD based method; estimated covariance matrix; harmonic retrieval; learning capacity; network training; noise-reduction neural network; preprocessing stage; singular value decomposition; sinusoids frequency estimation; state-space approach; white noise; Computational modeling; Covariance matrix; Feedforward neural networks; Frequency estimation; Intelligent networks; Neural networks; Noise reduction; Singular value decomposition; State-space methods; White noise;

fLanguage

English

Publisher

ieee

Conference_Titel

Circuits and Systems, 1990., IEEE International Symposium on

Conference_Location

New Orleans, LA

Type

conf

DOI

10.1109/ISCAS.1990.112093

Filename

112093