Research On Phase Match Methods Of Signal Denoising

With the rapid development of signal processing technology, it has been an essential task to remove the adverse effects of noise signal and to obtain the desired signal, which has been a constantly research topic for recent decades. An effective denoising method can be the crucial tool to remove noise interference and to reflect the original signal more realistically, which will make the signal analysis and information acquisition more convenient. Therefore, this paper gives five methods for signal denoising based on phase match methods.First, a wavelet denoising method based on noise estimating phase match is proposed. The signal with noise is processed for phase match in complex domain. Then the variance of the noise is acquired from mathematical derivation and the wavelet threshold is obtained by the noise variance. Though the noise information is not clearly known, the method can reflect the real-time changes of noise. The mathematics proof is done for this method. According to the simulation and comparison of the-3dB and-9dB SNR of the sine wave, square wave and triangle wave, the contrast shows that the denoising effect of the proposed method can been improved by4times to8times better than the Dohoho method in lower SNR.Second, a wavelet denoising method based on the line transition threshold is put forward. The range of threshold can be further extended when taking the soft and hard threshold wavelet denoising methods as special cases of this method. The grads of line transition threshold can be expediently adjusted according the actual situation to get different SNR signal, leading the output signal more smoothly. The simulation is under the situation that the signal is mixed with12dB White Gaussian noise of Block signal and Quadchirp. The gradient is3and Sym8three-layer wavelet is used. After the treatment of the proposed method, the signal amplitude is higher than that of the soft threshold denosing and the spikes jump change is lower than that of the hard threshold method. The SNR is improved and the signal becomes smoother. The standard deviation of denoised signal is lower than that of the soft and hard threshold denoising methods when the grads are1.5,3and6separately.Third, a denoising method for the high order spectrum based on the phase match of noise estimate is put forward. Apply the bispectrum denoising to the signal mixed with Gaussian and Non-Gaussian noise signal and make the two-dimensional Fourier transform to get the signal without Gausssian noise. Then estimate the noise spectrum by matching phase method. The spectrum difference denoising method is done in frequency domain to eliminate the Non-Gaussian noise. The proposed method gives the formula of how to obtain the noise spectrum and the spectrum difference in the paper. The simulation is under the condition that sine wave frequency is100Hz and the added noise is Non-Gaussian colored noise and White Gaussian noise with SNR-1dB,-5dB,-10dB separately. The simulation result shows that denoising effect is better than bispectrum method. The Gauss noise and non Gauss are both removed by this method,especially under the-10dB situation.Fourth, a denoising method based on phase match for the double array is proposed. The mathematical derivation of this method is given in this paper. The signal can be computed directly without mutation in the noise signal. The method can get more information and is conductive to counteract noise. The proposed method makes up the shortfall that the phase match method proposed by Orris which needs noise information priorly while the proposed method in this paper needs not. The experiment is under the condition that the sine signal frequency is300Hz and the SNR of While Gaussian noise are-70dB,-50dB and-30dB separately. With the sampling frequency is10kHz, a fitting degree table between the obtained signal and the original signal is given in this paper. The fitting degree is over90%. Besides, it also analyzes the influencet when noise amplitude and phase change. The conclusion is that there is no influence on the obtained result when the noise phase changes, and the coefficient K can be used to adjust the denoising results when noise amplitude changes.The match coefficient k can be used to adjust results when the signal phase changes.Under the condition that SNR is-50dB and the noise amplitude and noise phase are fluctuating, The fitting degree is over90%when simulation is done in different time periods and different K.Five, a method of separating signal from chaos background based on phase match is put forward. The signal can be obtained from the chaos background directly using the phase match method. This method does not need to extract signal using nonlinear minimization under minimum phase space. So it overcomes the shortcomings of poor separation under the low-dimensional flow-shaped state. The simulation is under the condition that chaos background is Lorenz and different frequencies and different phase sine waves are mixed.The separation table shows that signal is separated from the chaos background. The separation degree is low when the frequency is low and the amplitude is large. The separation degree is lower than3. The signal can been separated from the chaos background.At last, the core phase match method is applied to FPGA in order to design the soft IP core. The IP core is made up of a few models.The function simulation and RTL simulation are done in every model.The simulation results are consitent with the theoretical results, which lays the foundation for the application in embeded equipments.