Research On Power Quality Analysis System Base On Fast S-transform And Extend Kalman Filter

With the development of the power grid and the progress of the society in China, increasing amount of renewable and distributed energy, high-power electronic devices are connected to the grid, which brings unprecedented challenges to power quality. More and more attention is paid to power quality problem. High power supply reliability and power quality are mentioned as core responsibilities of the State Grid of China. Thus it is significant to do researches on power quality in depth.The study of power quality issue needs some effective algorithms to extract the feature vectors of power disturbances. This paper introduced the fast S-transform which improves the classical S-transform. Compared to the traditional S-transform, it can eliminate the redundancy of the analysis result, reduce the computational complexity and decrease the storage space. The computational complexity of fast S-transform can match with which of fast Fourier transform, so it can be used in real-time analysis. By comparing the analysis results of the power disturbances using the fast S transform and wavelet transform theory, this paper has verified that the fast S-transform can extract feature vectors from power disturbances effectively. The results also shows that the fast S-transform works better than wavelet transform when dealing with noise.In order to enhance the speed, stability and accuracy of power signal parameter calculation, this paper proposed an improved algorithm which combines the traditional Extended Kalman Filter with the fast S-transform. By using the feature vectors from Stransform, the function of Extended Kalman Filter is strengthened. Based on the MATLAB platform, this paper verified that the method can enhance the power signal parameter calculation ability when signal suddenly changes or noise exists in the signal.Finally, a system consists of a DSP and a computer server is constructed to test the above method. The DSP can sample power quality signals from an arbitrary waveform generator and upload the data to computer using TCP protocol. Then the data is analyzed on the computer server using the algorithm mentioned above to extract feature vectors and calculate signal parameters. The experiment results showed that the method can realize real-time analysis of power quality signals effectively.