随机信号离散模型与连续模型等效转换的研究

本文提出一种应用E变换法对平稳随机信号的ARMA离散模型与AM连续模型等效转换的新方法；给出它与已有基于格林函数等效转换法的关系；对此种新方法应用于非平衡随机信号的ARMA时变参数离散模型与AM时变参数连续模型的等效转换做了初步探讨。     

非平稳信号的一种ARMA模型参数估计法

本文采用一种经过特殊处理的时变自回归滑动平均（ARMA）模型对非平稳随机信号进行分析．将这种模型左边的时变参数假设为一组基时间函数的线性组合，右边时变参数简化为常数，并用反馈线性估计法进行参数估计。该方法的主要特点是简单，计算量小，占用存储空间少．并用仿真的方法对算法予以验证，可用于一些常用的非平稳随机信号的分析．     

非平稳信号分析的广义解析模态分解方法

现有的非平稳信号分析方法都有各自不同的缺陷,短时傅里叶变换的时频分辨率受不确定性原理的限制,希尔伯特黄变换存在端点效应和模态混叠,易导致模糊的时频分布;解析模态分解只适合分析频率恒定的多分量信号;针对包含多个时变模态、特别是频谱重叠的非平稳信号,本文提出了一种新的信号分析方法———广义解析模态分解（Generalized Analytical Mode Decomposition,GAMD）．GAMD通过广义傅里叶变换将时变频率转换为频谱可分的,采用解析模态分解对其分解,再对得到的单分量信号进行逆广义傅里叶变换即可得到原始信号的分量．因此,GAMD非常适合分析时变的非平稳信号．通过仿真信号将GAMD与短时傅里叶变换和希尔伯特黄变换等方法进行了对比,结果表明GAMD方法的分解效果更精确,时频分辨率更高．     

一种高精度的非平稳随机信号实时建模算法研究

针对传统的非平稳随机信号时变参数筹分模型算法的不足,论文把非平稳随机信号时变参数差分模型与一种带自适应遗忘因子的RLS算法结合起来,形成一种有效的非平稳随机信号实时建模算法,仿真分析表明该算法不仅能获得快的收敛速度而且能获得高的建模精度。     

基于经验模式分解的Wigner-Ville分布交叉项抑制方法

时频分析作为时变非平稳信号分析的有力工具,成为现代信号处理研究的一个热点.这种分析方法提供了时间域与频率域的联合分布信息,为我们清楚地描述了信号随时间变化的关系.Wigner-Ville分布由于其良好时频集聚性,在非平稳信号分析中得到广泛应用,本文针对Wigner-Ville分布中的交叉项问题,提出了基于经验模式分解的Wigner-Ville分布,即对多分量信号运用经验模式分解,将其分解为单分量信号,再对每个单分量信号求Wigner-Ville分布进行线性叠加.提出运用相关系数法对经验模式分解伪分量进行剔除,提高了该方法的精度,并将该方法与Cohen类方法进行比较,阐述了该方法的优点.     

融合EMD与全局版人工鱼群LS-SVM的网络流量预测

为提高网络流量预测模型对稳定性和精度要求,提出了一种融合经验模式分解(EMD)和全局版人工鱼群LS-SVM模型的网络流量预测算法。该模型首先通过EMD分解,将原始流量的非平稳序列分解为若干不同尺度的平稳IMF(固有模态分量)分量,再用LS-SVM模型分别对各分量进行预测,其中LS-SVM的参数用全局版人工鱼群算法进行优化;最后通过SVM组合得到原始序列的预测值。仿真实验验证了该算法的预测准确性和稳定性。     

基于EMD的语音功率谱分析

提出了一种基于EMD的功率谱分析法,即先用经验模态分解方法(EMD)将语音信号分解成若干个固有模态函数IMF分量,而后对包含主要信息的IMF分量利用现代参数模型法估计出它们各自的功率谱。文中对不同情感状态下语音数据的分析结果表明,EMD方法能有效地应用于非平稳语音信号的功率谱分析中,可更细致的体现语音信号内在特征。     

LMD结合时变AR模型在瞬时频率测量中的应用

时变AR模型是一种新的时频分析方法,对简单的非平稳信号瞬时频率的测量具有很高的精度,但却不适用于复杂信号。为使时变AR模型同样适合于测量复杂信号的瞬时频率,提出LMD和时变AR模型相结合的方法,该方法使时变AR模型极大地扩展了适用对象。分别采用最小二乘和卡尔曼平滑算法求解信号参数模型,并求得信号瞬时频率,仿真对比分析表明,与直接用时变AR模型测量瞬时频率相比,该改进方法能大大提升瞬时频率测量的准确度。     

基于蒙特卡洛滤波平滑的语音增强算法

本文在分析统计信号贝叶斯模型和语音信号的时变自回归(TVAR)模型的基础上,利用蒙特卡洛滤波及平滑方法,对语音信号的TVAR模型参数进行了估计,提出了一种有效的针对非平稳加性噪声影响下的语音增强算法.该算法可以很好的跟踪非平稳信号,同时引入对反射系数的判断,保证了跟踪的稳定性.实验表明,本文方法能很好的抑制背景噪声,提高信噪比,改善语音信号的听觉质量.     

利用自适应傅里叶分解的非平稳无线信道的时频表示

针对非平稳Clarke无线信道模型的时域冲激响应的性能分析需求,利用自适应傅里叶分解,引入了一种非平稳无线信道的时频表示方法和信道函数的重构表示,并给出了信道的单分量表示式、时间频率分布以及能量谱密度。在高速移动、快速时变环境下进行仿真,结果表明,本文提出的非平稳无线信道的表示方法能克服STFT、小波变换等相关方法的缺点,提高了无线信道时频表示的准确性,降低了信道的重构误差。      

谱跟踪和噪声模型语音信号分析/合成方法

把语音信号分解为确定和随机两部分是非常典型的语音信号分析与合成方法。针对目前主流研究的谐波加噪声模型(Harmonic plus Noise Model, HNM)和准谐波模型(Quasi-Harmonic Model, QHM)法等存在冗余大的问题,提出了一种基于谱跟踪和噪声模型的语音信号分析/合成方法。以音素为基本单元进行编解码,用谱跟踪法提取时变特征合成语音信号的确定部分,用AR模型合成语音信号的随机部分。实验发现,合成语音与原语音波形上比较接近,MOS分约为3.24,冗余也明显减少。     

Fetal Electrocardiogram Enhancement by Time-Sequenced Adaptive Filtering

An adaptive method for performing optimal time-varying filtering of nonstationary signals having a recurring statistical character, e.g., recurring pulses in noise, has been proposed. This method, called time-sequenced adaptive filtering, is applied to the enhancement of abdominally derived fetal electrocardiograms against background muscle noise. It is shown that substantial improvement in terms of signal distortion is obtained when time-sequenced filtering, rather than conventional time-invariant filtering, is employed. The method requires two or more abdominal channels containing correlated signal components, but uncorrelated muscle noise components. The location of the fetal pulses in time must be estimated in order to synchronize the filter's time-varying impulse response to the fetal cardiac cycle.     

时频联合处理方法在地波高频雷达中的应用

一般采用时域或频域的方法对雷达信号进行分析。时频联合处理方法在时间-频率二维空间对信号的特性进行分析,从而进一步得到信号频谱分量在时间轴上的分布图。因此,对于具有时变频率分量的空中目标回波而言,时频联合处理方法是一种有效的分析方法。文中简要介绍了时频联合处理方法的基本方程,采用时频联合处理方法对具有多个频谱分量的混合信号、噪声加信号和高频雷达在强海杂波背景下的空中机动目标回波信号进行了分析和说明。结果表明,时频联合处理方法的应用能大大提高高频雷达对目标的探测能力。     

FMmlet变换在信号分离中的应用

具有非线性时频关系的多分量信号的频谱占有较宽的频带，并且各信号分量的频谱常常互相交叠，因之难以用常规的滤波方法将各信号分量分离，将信号与背景噪声分离亦非易事，由于FM^mlet变换的原子能匹配信号的线性和非线性时变成分，因此本文提出采用FM^mlet变换进行信号分离，实际算例结果印证了本文方法的有效性。     

Time-frequency analysis of HRV for patient with vasovagal syncope

Time-frequency analysis (TFA) is applied to the heart rate variability (HRV) signal for the purpose of observing the dynamical changes of spectral components. Smoothed-pseudo Wigner-Ville distribution (SPWVD) TFA was considered for processing the HRV signal. As a result, SPWVD TFA on HRV signals is shown to be a very effective method for analysing the power spectral density and for observing the time-varying components of HRV signals     

Regularized Spectral Matching for Blind Source Separation. Application to fMRI Imaging

The main contribution of this paper is to present a Bayesian approach for solving the noisy instantaneous blind source separation problem based on second-order statistics of the time-varying spectrum. The success of the blind estimation relies on the nonstationarity of the second-order statistics and their intersource diversity. Choosing the time-frequency domain as the signal representation space and transforming the data by a short-time Fourier transform (STFT), our method presents a simple EM algorithm that can efficiently deal with the time-varying spectrum diversity of the sources. The estimation variance of the STFT is reduced by averaging across time-frequency subdomains. The algorithm is demonstrated on a standard functional resonance imaging (fMRI) experiment involving visual stimuli in a block design. Explicitly taking into account the noise in the model, the proposed algorithm has the advantage of extracting only relevant task-related components and considers the remaining components (artifacts) to be noise.     

Adaptive Complex-Coefficient 2D FIR Trapezoidal Filters for Broadband Beamforming in Cognitive Radio Systems

A novel design method is proposed for an adaptive discrete-domain beamformer for the beamforming of temporally broadband-bandpass signals in cognitive radio (CR) systems. The method is based on a complex-coefficient 2D finite impulse response (FIR) filter having a trapezoidal-shaped passband. The temporally broadband-bandpass signals are received by a 1D uniformly distributed antenna array (1D UDAA), where the outputs of the antennas are complex-quadrature sampled by the front end of the CR system. This CR system is based on a software defined radio (SDR) architecture and can be instantly reconfigured by the control system to select the appropriate frequency band and the required sampling rate. The subsequent beamforming enhances the spectral components of the desired temporally broadband-bandpass signals by arranging for the asymmetric trapezoidal-shaped passband of the 2D filter transfer function to closely enclose the region of support (ROS) of the spectrum of the desired signal, whereas the ROSs of the spectral components of the interfering signals are enclosed by the stopband. The proposed novel closed-form design method facilitates instant adaptation of the shape and orientation of the passband of the beamforming 2D FIR trapezoidal filter in order to match the time-varying frequency band and the time-varying bandwidth of the signal, as well as to track and enhance received signals with time-varying directions of arrival (DOAs). Simulated results confirm that, compared with previously reported methods, the proposed method achieves the best overall tradeoff with respect to the instantaneous adaptations of the operating frequency band, the bandwidth, and the time-varying DOAs, the distortion of the desired passband signal, and the stopband attenuation of interfering signals.     

Performance of Phase-Locked Loops in the Presence of Fading Communication Channels

An approach is presented for the analysis of phase-locked loops whose input signal has passed through time-varying channels. The specific channels considered in detail are the Rice-Nakagami, Rayleigh, and lognormal fading channels. Loop performance is characterized in terms of the steady-state probability density function of the reduced phase error process. The basic parameters which characterize performance include the loop signal-to-noise ratio (SNR) and the variance and bandwidth of the fading components introduced by the channel. Particular channel models are used to illustrate the theory for the firstorder loop. The results are also applied to the analysis of the PSK noisy reference problem in the presence of these time-varying channels.     

基于变参数随机共振和归一化变换的时变信号检测与恢复

非线性随机共振系统具有利用噪声增强微弱信号的能力,为强噪声背景下的信号检测开辟了新的途径。该文提出一种变参数随机共振(VPSR)模型,实现对非周期信号的有效检测、噪声去除和信号恢复。通过以恢复信号的拟合决定系数和互相关系数作为评判标准,研究分析了不同参数变化对系统输出的影响,分析结果表明该模型能有效地从噪声背景中恢复时变信号。该方法拓展了随机共振用于时变信号检测技术的领域,在时变信号检测和处理以及雷达通讯等方向有着一定的潜在应用价值。     

脑电信号成分稀疏分析范式及其可行证明

稀疏分解技术作为一种可靠的信号处理与传输方法,在包括EEG的多种时变信号分析和处理领域得到了广泛的应用。在EEG信号的成分分析中,现有算法（ICA,EMD）等都存在分解结果与真实成分显著不符的情况,难以对实际成分的波形进行估计。本文在稀疏分解算法基础上,通过对样本稀疏分布情况进行度量,给出了一个经过改良的稀疏性能评价指标（SPI）并以此建立了一个新的成分分析范式和相应的优化函数,经过理论和实际证明,该范式在成分分析领域能比传统方法更有效地使分解结果趋向于真实成分,对EEG信号、乃至其他时变信号的成分解析都具有相当的积极意义。