基于广义S变换的时频滤波技术研究
|
摘要
针对S变换的滤波方法受到窗函数固定的限制,对信号的滤波不能达到良好效果,广义S变换时频滤波克服了传统S变换滤波因子不能随时间、频率变化而变化的缺陷。将信号用广义S变换方法变换到时频域,对不同时间内不同频率的噪声部分冲零,再将去噪后的信号利用广义S反变换到时间域,获得所需要的有效信号。通过理论计算和信号模型仿真表明,广义S变换时频滤波方法能够较为精确的分析数据的时间和频率特征,有效滤除不同时段不同频率的噪声,可以最大化的保留原始信息。该方法具有较高的实用性和灵活性。
S-transform filtering methods are restricted because window function is fixed,as a result,a satisfactory effect of filtering can not be obtained from actual data.Time-frequency filtering with Generalized S-transform overcomes the shortage that the filtering factors in traditional S-transform filtering and denoising approaches will not change with time and frequency variation.Signal data are transformed to time-frequency domain by using the Generalized S-transformation methods,then,the noises at different time intervals and with different frequencies are zeroed partly.Finally,the signal data after noise elimination are transformed into time domain again by using Generalized S-inversetransformation to achieve the effective signals needed.Based on theoretical and real signal model simulation,it is believed that Generalized S-transform is really effective method eliminating noises at different time intervals and with different frequencies.Generalized Stransformcould reserve the primary information maximally and accurately analyze the characteristics of the data in timefrequency domain.This method possesses adjustable time-frequency resolution with higher practicability and adaptability than other filtering methods.
S-transform filtering methods are restricted because window function is fixed,as a result,a satisfactory effect of filtering can not be obtained from actual data.Time-frequency filtering with Generalized S-transform overcomes the shortage that the filtering factors in traditional S-transform filtering and denoising approaches will not change with time and frequency variation.Signal data are transformed to time-frequency domain by using the Generalized S-transformation methods,then,the noises at different time intervals and with different frequencies are zeroed partly.Finally,the signal data after noise elimination are transformed into time domain again by using Generalized S-inversetransformation to achieve the effective signals needed.Based on theoretical and real signal model simulation,it is believed that Generalized S-transform is really effective method eliminating noises at different time intervals and with different frequencies.Generalized Stransformcould reserve the primary information maximally and accurately analyze the characteristics of the data in timefrequency domain.This method possesses adjustable time-frequency resolution with higher practicability and adaptability than other filtering methods.
引文
[1]STOCKWELL R G,MANSINHA L,LOWE R P.Location of the complex spectrum;the Stransform[J],IEEETransactions on Signal Processing,1996;44(4):998-1001.
[2]高静怀,陈文超,李幼铭等广义S变换与地层地震响应分析[j].地球物理学报.2003,46(4):526-532.
[3]刘喜武,刘洪,李幼铭等基于广义S变换的地震波能量衰减[j].勘探地球物理进展,2006.29(1):20-24.
[4]PINNEGAR C R,MANSINHA L.The Stransformwith windows of arbitrary and varying shape[J].Geophys-Ics,2003;68(1):381-385.
[5]PINNEGAR C R,MANSINHA L.Time-local spec-tral analysis for non-stationary time series:the S-trans-form for noisy signals[J].Fluctuation and Noise Letters,2003;3(3):357-364.
[2]高静怀,陈文超,李幼铭等广义S变换与地层地震响应分析[j].地球物理学报.2003,46(4):526-532.
[3]刘喜武,刘洪,李幼铭等基于广义S变换的地震波能量衰减[j].勘探地球物理进展,2006.29(1):20-24.
[4]PINNEGAR C R,MANSINHA L.The Stransformwith windows of arbitrary and varying shape[J].Geophys-Ics,2003;68(1):381-385.
[5]PINNEGAR C R,MANSINHA L.Time-local spec-tral analysis for non-stationary time series:the S-trans-form for noisy signals[J].Fluctuation and Noise Letters,2003;3(3):357-364.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心