Kurtosis Emd Pulsar Signal Denoising Method For Testing Window

Pulsars are a class of neutron stars with strong magnetic fields and rapid rotation.The most notable characteristic of pulsars is a very stable signal radiation cycle,this feature show a very large potential applications in the pulsar navigation,radio communication,astrophysics and other areas.Due to the interstellar space material interference and long distance as well as receiving physical equipment performance and other factors,the received pulse signal contains relatively strong noise.But the signal noise ratio(SNR)of pulsar signals is proportional with navigation precision.Therefore,getting a higher SNR under a low SNR of the pulsar signal de-noising is a premise basis in pulsar observations and other practical applications.The content of this paper is to study pulsar de-noising algorithm.The common Pulsar signal de-noising methods are commonly used Fourier transform method,wavelet analysis,empirical mode decomposition and so on.Since the Fourier transform method is only suitable for stationary signals,but the pulsar signal is a non-stationary signal,so this method does not apply to the pulsar signal.The wavelets and optimal decomposition level of wavelet analysis method are difficult to choice.Therefore,this paper chooses empirical mode decomposition(EMD)algorithm,because EMD needn't to pre-determine the substrate and the decomposition level,it determines its substrate and the decomposition level by their own.On top of this,this paper presents a window function method based on the EMD.About the aliasing of intrinsic mode function(intrinsic mode function,IMF)between noise and useful signal in the EMD decomposition process.Firstly,by autocorrelation and cross correlation calculation,it finds the starting IMF of signal.Secondly,it determines the point between noise and useful signal in the IMF followed by partial kurtosis test method,it uses the Turkey-Hanning window to retain the useful signal and inhibitions the noise signal.Finally,it uses the Adaptive thresholding to achieve separation of signal and noise and to improve the quality of the reconstructed signal.The simulation and experimental results show that compared with the adaptive threshold and window function method,the paper de-noising method can more effectively filter the noise in the signal the signal pulse,while better keep the details of the signal characteristic in the mutation position,so it has a higher noise cancellation performance.