Parameter Correcting Method For Discrete Spectrum With Few Sampling Points

Discrete spectrum correction technology is widely used in mechanical equipment fault diagnosis,rotor dynamic balance,vibration analysis,automatic control processes,image processing,language recognition,petroleum exploration,marine resource survey,and biomedical engineering,etc.Due to the energy leakage phenomenon and the fence effect of the discrete spectrum,the frequency,amplitude and phase parameters of the discrete spectrum obtained by the fast Fourier transform often produce large errors.Therefore,it is of great significance to study and improve the accuracy of discrete spectrum correction technology.Because discrete spectrum correction methods such as phase difference correction method have significantly reduced correction accuracy when the number of initial signal points is small,this paper first proposes a method to improve the correction accuracy of phase difference correction method by using different interpolation methods.In this method,linear interpolation or cubic spline interpolation is performed on an existing sampling signal to obtain more sampling points,thereby improving correction accuracy.And the two important parameters in the phase difference correction method are studied in detail.Simulation results show that when the initial sampling points are few and noisy,this method can improve the accuracy of the phase difference correction method to a certain extent.This method can be used to initially determine the frequency range.Secondly,the application of zero-padding method in discrete spectrum parameter correction is studied.First,the application of the zero-padded method when the number of supplementary data points and the number of initial data points are small is studied.Then,a ratio correction method combined with the zero-padded method is derived.The normalized frequency correction amount is obtained by the peak search method,and the amplitude and Phase correction formula.Finally,the simulation results in Matlab are used to study the correction results when different window functions are added with or without noise.