Parameter Estimation And Accuracy Analysis For AM Sinusoidal Signal

As the demands for the intelligent communications or real-time multimedia continue to increase, the types of processing signal are not only limited to conventional constant parameter signals, but also include a large number of changing parameter signals. Amplitude modulated sinusoidal signal which is widely used in engineering practice, voice appreciation and other areas is a common non-stationary signal. Nowadays, changing parameter signal parameter estimation in noise background has a wide range of applications in automatic control, vibration signal processing and plasma processing, music appreciation and other fields, and has become an important analytical tool.This thesis focuses on the research of parameter estimation and accuracy analysis for the amplitude modulated sinusoidal signal. Firstly, the short-time DFT algorithm is used to estimate the instantaneous phase, amplitude and power for the AM signal in the field of pulse power discharge, then their expressions are derived and simplified, respectively. In order to treat the spectral leakage problem, the influence of Hanning window is studied and the accuracy of parameter estimation impacted by AM is analyzed on the basis of the simplified expression. Computer simulation results verify that the derivation expressions are correct. Then, it gives the conditions of AM signal frequency estimation by using amplitude interpolation estimation algorithm based on the Fourier transform, and using the signal frequency shift and applying the Hanning window method and carries on the corresponding improvement on the basis of existing amplitude interpolation algorithm. In addition, the performance of each estimation algorithm is analyzed under the conditions of noise and the amplitude modulation impact on the estimated result. Finally, the multi-point estimation method and curve fitting method are used to estimate the short waveform AM sinusoidal signal which is less than one period in time domain. Multi-function point estimation method makes the appropriate amplitude combination by using the sine characteristics to derive signal frequency. The curve fitting method applies the least squares method to obtain an approximate polynomial expression of AM signal, and uses the coefficients of Taylor expansion corresponding to approximate the polynomial coefficients to obtain an estimation of the signal frequency, The simulation results show that the method can effectively estimate the frequency of AM sinusoidal signal.