Research On Harmonic Detection Method Based On Hybrid Convolution Window And Compressed Sensing

With the development of power system and the popularization of new energy sources,modern power electronic equipments have been widely used,and many nonlinear,impulsive and unbalanced loads are connected to the power grid.The consequent harmonic pollution problem is gradually aggravated,threatening the security and stability of the power grid.In order to control the influence of harmonic pollution,it is necessary to measure the harmonic signal parameters accurately and effectively,and designing a harmonic detection method with excellent performance is the key.Based on the mature FFT harmonic detection algorithm,this paper studies a new harmonic detection method,which is improved from two aspects: reducing error and reducing sampling data storage.Since it is difficult to achieve ideal synchronous sampling in the actual sampling process,FFT detection algorithm will have spectrum leakage and fence effect.In view of the above two problems,it is more effective to cut off the signal by windowing and correct the detection data by spectral line interpolation.The detection accuracy of windowed interpolation algorithm is affected by the ability of window function to suppress spectral leakage,which depends on the spectral characteristics of window function.Generally,the sidelobe characteristics of cosine window are not ideal.Therefore,the concept of convolution is introduced,and the different window functions are convolved to obtain a hybrid convolution window with better sidelobe performance.Compared with ordinary cosine window and partial self-convolution window,it has stronger ability to suppress spectrum leakage and higher detection accuracy.On this basis,a FFT harmonic detection algorithm based on 4-term 5-order Nuttall and 5-term MSLD hybrid convolution window is proposed.Experiments show that its detection performance is better than other hybrid convolution windows,and it can effectively overcome the influence of frequency fluctuation and white noise interference.Compressed sensing technology is not limited by Nyquist sampling theorem,it can greatly reduce the amount of sampled data and save storage space.SAMP reconstruction algorithm can effectively reconstruct signals with unknown sparsity,which is very suitable for harmonic detection.Therefore,by combining the hybrid convolution window with the compressed sensing theory,a four-spectrum-line SAMP algorithm with 4-term 5-order Nuttall and 5-term MSLD hybrid convolution window is proposed,which can not only ensure the accuracy of the detection algorithm,but also use compressed sensing to greatly reduce the sampling amount and save the storage space.Experiments show that the harmonic detection method based on hybrid convolution window and compressed sensing has a relative decline in the accuracy of harmonic detection when the sampling data is only 1/4 of the traditional FFT algorithm,but it is still far higher than the requirements of the national standard for the accuracy of harmonic detection equipment.This thesis includes 42 figures,15 tables and 96 references.