功率谱计算(国外英文资料).docVIP

功率谱计算(国外英文资料) Power spectrum estimation is a very important topic in modern signal processing, and involves many problems. Here, in conjunction with MATLAB, let me make a rough introduction. Power spectrum estimation can be divided into classical spectrum estimation method and modern spectral estimation method. The simplest method in classical spectral estimation is periodogram, which is divided into direct and indirect methods. Fu Liye transform N direct method first point data (i.e. spectrum), and then take the product spectrum conjugate, get the power spectrum estimation; indirect method to calculate the autocorrelation function of N sample data, and then from the Fu Liye transform of the correlation function, power spectrum estimation is obtained. It can be realized simple. In MATLAB, periodogram can be implemented with function periodogram. However, the power spectrum estimated by periodogram is not fine enough and the resolution is low. So it is necessary to revise the periodogram, can signal sequence x (n) n is divided into non overlapping segments, respectively with the periodogram spectral estimation, and then the average value of the N data to estimate the results as a whole data of power spectrum estimation results. The signal sequence x (n) can be overlapped and segmented to calculate the power spectrum respectively, and then the average value is calculated as the power spectrum estimation of the whole data. These 2 are called piecewise mean periodogram, and the latter is generally better than the former. The windowed mean periodogram method is an improvement to the piecewise mean periodogram method, that is, after the data is segmented, a rectangular window is added to each section for preprocessing, and then the power spectrum is estimated by the piecewise mean periodogram method. Compared with the block average periodogram method, the windowed average periodogram method can reduce the frequency leakage and increase the frequency peak width. Welch meth