Abstract: The major parameter estimation error source for discrete Fourier transform are spectrum leakage and picket fence effect. A weighted two points vector interpolation algorithm based on Rife-Vincent window for improving the estimation accuracy of parameters for the multi-frequency damped signals is presented, where the signals are weighted by M-order cosine window family before the discrete Fourier transform and the equations with nuknown of frequency deviation and decay factor is established by the ration of two adjacent spectral lines near the real frequency position. The frequencies, amplitudes and phases are calculated by the frequency deviation solved from the above equation. The cosine window family with the characteristic of maximum sidelobe decay can essentially remove the spectral leakage and two point vector interpolation will cope with the problem of the picket fence. To combine the merits of the above two method can effectively increase the accuracy. The simulation and test results show that the proposed algorithm has a higher estimate accuracy and steady estimates. The higher computational efficiency and lower memory demand are suggested especially for poor computing resource situations. It also can be used as an optional method for the features extraction of the multi-frequency damped signals.

Key words: frequency domain interpolation, parameter estimation, Rife-Vincent window, spectral leakage