A SVD-based Adaptive Kalman Filter For Detection Of Harmonics And Voltage Sag

With the rapid growth of economics, power quality issue has increasingly captured considerable attention from both utility companies and their customers. On one hand, with the increase of nonlinear loads, there are more and more factors that will bring challenge on the power quality, for instance the progress on power electronics appliance and its implementation in industry and transportation companies. On the other hand, people have higher request on the power quality and reliability of power system because more complex, rigid and power quality-sensitive appliance is widely used.Power harmonics and voltage sag are two important matters in power quality research. The research key to them has been from quantities analysis to put them in order now, namely the research to compensation equipment. A good detection way is absolutely necessarily to compensation equipment for getting excellent effect. Thereby, this paper has studied the detection of harmonics and voltage sag.First of all, usual detections of harmonics and voltage sag are analysed in the paper. The deficiency of usual detections has been pointed out one by one through theory analysis and simulation. The kalman Filter can be used to detect harmonics and voltage sag in virtue of fairly high speediness and precision of detection. However, the traditional kalman filter has some deficiencies. Firstly, the filter divergence will happen because of inexact estimation to statistic characteristic of the noise and rounding error of computer. Secondly, when kalman filter has reachedsteady state its error covariance will have a very small stationary value, namely "filter dropping off'. Thirdly, the traditional kalman filter cannot adjust its filter parameters with statistic characteristic of the noise. To overcome the deficiency of kalman filter, an SVD-based adaptive kalman filter is put forward in the paper. The filter restrains the filter divergence by matrix singular value decomposition. The eigenvalues and eigenvectors of the matrix are insensitive to tiny change of matrix element and acceptable error. Then, a transient indication function can be constructed to judge whether abrupt changes of signal has happened. Once a transient has been detected kalman filter reset the value of error covariance matrix to a predefined high level so as to increase the sensitivity of kalman filter and match the changes of the signal. Finally, kalman filter has been made possess strong adaptive ability by improving Sage-Husa algorithm. The validity has been proved through simulation analysis. At the same time its detection of harmonics and voltage sag has excellent performance.