Estimation and filtering of harmonics.

Abstract

The extensive use of modern nonlinear power electronic devices and nonlinear loads has made the problem of harmonic distortion more challenging to researchers and utility engineers. Estimation of these serious harmonic interferences is necessary for the analysis and design of modern equipment and for the system security. The need for the effective elimination of such unwanted frequency components becomes essential in power system planning, analysis, and operation. The work proposed in this thesis is two folds. In the first part, different frequency components of a distorted signal have been estimated. Different estimation algorithms were used and tested over standard signals with different frequency components present. The robustness of the algorithms was tested in the context of additive noise, for different noise levels, corresponding to different signal to noise ratios (SNR). In the second part of the thesis, two new filtering schemes for harmonic elimination are proposed. These techniques have been studied for different types of practical signals reported in the literature. The robustness of the schemes has been tested in context with the additive noise of different SNRs. Apart from the simulation of these techniques, the prototype implementations on a small voltage scale are also presented and evaluated based upon the criteria of robustness, accuracy and reliability.