Research On Harmonic Analysis Of Power System Based On Cosine Self-convolution Window Improved FFT

Accurate detection of harmonic parameters has become the premise and key to reduce harmonic pollution,prevent the harm of harmonics and maintain the safe operation of power grid.Fast Fourier transform(FFT)algorithm has become a typical harmonic detection method because of its simple principle and easy realization in hardware circuit.In the case of non synchronous sampling,it is easy to produce the spectrum leakage and fence effect by applying FFT directly to the harmonic analysis.The window function is used to solve the spectrum leakage in time domain,and the interpolation of the discrete spectrum can reduce the error caused by the fence effect.Aiming at the existing problems of FFT harmonic detection,this paper analyzes the frequency spectrum characteristics of the classical window in order to lay the foundation for constructing the new window function and the improved FFT power harmonic analysis method.The main work of this paper:(1)Based on the analysis of cosine window function,four-item five-order Nuttall window and five-item maximum-sidelobe-decay window with the best sidelobe performance being choose as the parent window are used to construct the self-convolution window.Compared with the classical cosine window,cosine self-convolution window sidelobe performance is more superior,to further suppress the spectral leakage,and with the increase of the convolution order,the rapid reduction of the sidelobe peak level and sidelobe attenuation rate of the inhibition spectrum leakage ability rapid enhancement.At the same time,the four-spectral-line FFT interpolation algorithm based on the five-item MSD convolution window is constructed by using the four spectral lines near the frequency points.Phase difference FFT algorithm based on four-item five-order Nuttall self-convolution window is constructed by using phase difference algorithm.(2)This paper introduces the measurement principle of dielectric loss angle,and introduces consine hybrid-convolution window triple-spectral-line interpolation FFT and the five-item MSD window improved four-spectral-line FFT to improve the accuracy of dielectric loss angle measurement.The interpolation polynomial of the three-spectral-line and the four spectral lines are derived by means of polynomial fitting.At the same time using any adjacent main lobe in two line phase difference ?(or-?)characteristics,reduce the four-spectral-line interpolation parameter detection in the process of the operation mode for the times,to a certain extent reduces the amount of computation.The simulation results show that this chapter proposed consine hybrid-convolution window triple-spectral-line interpolation FFT and the five-item MSD self-convolution window improved four-spectral-line FFT in the detection of dielectric loss angle,compared with the conventional windowed interpolation algorithm,the accuracy of detection of dielectric loss is still very high,and compared with the conventional algorithm,to a certain extent reduces the amount of computation.(3)On the basis of learning FFT_IP core and Verilog syntax,drive FFT_IP core to complete FFT operation based on FPGA,and lay the foundation for achieving windowed interpolation FFT algorithm based on FPGA.Finally,the accuracy and precision of the FFT algorithm based on the FFT_IP core in FPGA is verified by the two simulation experiments of synchronous whole period sampling and asynchronous non integer period sampling.