Hanning窗在电力系统谐波分析中的应用

采用FFT算法对电网信号进行谐波分析时很难做到同步采样和整数周期截断,由此造成的频谱泄漏和栅栏效应将影响到谐波分析的结果。本文应用矩形窗和Hanning窗的加窗插值FFT算法分析非同步采样的电力系统谐波,经过MATLAB仿真证明：采用基于Hanning窗的加窗插值FFT算法能够大幅度降低由非同步采样造成的误差,最后给出了实现该算法的C语言程序。     

高分辨图像二维FFT正/反变换实时处理方法及硬件实现*

高精度密集型数值计算和大规模数据缓存,是高分辨率图像二维FFT(快速傅里叶变换)实时实现中的主要难点。利用实信号傅里叶变换的周期对称性和频域数据的共轭对称性,提出了一种高效且易于硬件实现的二维FFT正/反变换的实时处理方法,将实值图像二维FFT中的一维FFT计算和存储需求缩减了近一半。在以4片TS201为计算核心的DSP处理平台上,使用该方法实现了二维FFT正/反变换和图像频域滤波。实验表明,无须片外存储,单片TS201可处理最大512×512像素的图像;该尺寸图像的正/反变换总处理时间为49.6 ms,     

Against the long-range spectral leakage of the cosine window family

Suppressing spectral leakage in the fast Fourier transform (FFT) has been investigated for over 30 years. Regarding the frequently used cosine window family, it is observed that the long-range leakage sampled by FFT spectral lines follow a flat trajectory. Consequently, the long-range leakage is approximated by polynomials in this paper. In light of this parametric model, the interpolating formula is presented with up to nine-point for a cosine window with maximum side lobe decaying. Its expression is general in the window order and number of interpolating points. Some well-known formulas of the modulus-based interpolated FFT are parallel to special cases of the new formula, but the former are susceptible to significant bias at coherent sampling conditions. The new formula was tested with real-valued signals containing a single tone and then duel tones. It is demonstrated the new formula is easy to implement and is free of the significant bias aforementioned.     

基于快速傅里叶变换的正弦信号频率高精度估计算法

为了进一步提高加性高斯白噪声背景中正弦信号的频率估计精度,提出了一种新的基于插值快速傅里叶变换(FFT)的正弦信号频率估计算法.首先,对N点正弦采样序列进行等长度时域补零延长,再进行 2N 点FFT; 然后, 搜索幅度最大离散谱线位置得到频率粗估计值; 最后, 采用幅度最大谱线以及原信号的离散时间傅里叶变换(DTFT)在幅度最大谱线左右两侧的两点抽样值进行精估计.仿真结果表明,当信号实际频率位于FFT两条离散谱线之间任意位置时,所提算法的频率估计均方根误差均接近克拉美罗下限,具有较好的一致性,估计精度高于Candan算法、Fang算法、三谱线合理结合(RCTSL)算法和Aboutanios算法, 且信噪比阈值较低,估计性能优于现有频率估计算法.     

基于FFT频谱校正的移频信号检测算法的研究

轨道电路移频信号是否正常直接影响到列车的行车安全。为了准确、高精度地检测轨道电路移频信号参数,利用汉宁窗的频谱校正和欠采样技术,给出了基于快速傅里叶变换(FFT)的频谱校正算法。利用Matlab对其进行仿真,结果表明各参数检测值均满足误差指标要求,且将采样时间缩短到1.6 s(或2 s)。该算法为研究实时移频信号参数测试仪表提供了理论依据。     

A novel fast nonstationary noise tracking approach based on MMSE spectral power estimator

Estimating the noise power spectral density (PSD) from the corrupted speech signal is an essential component for speech enhancement algorithms. In this paper, a novel noise PSD estimation algorithm based on minimum mean-square error (MMSE) is proposed. The noise PSD estimate is obtained by recursively smoothing the MMSE estimation of the current noise spectral power. For the noise spectral power estimation, a spectral weighting function is derived, which depends on the a priori signal-to-noise ratio (SNR). Since the speech spectral power is highly important for the a priori SNR estimate, this paper proposes an MMSE spectral power estimator incorporating speech presence uncertainty (SPU) for speech spectral power estimate to improve the a priori SNR estimate. Moreover, a bias correction factor is derived for speech spectral power estimation bias. Then, the estimated speech spectral power is used in “decision-directed” (DD) estimator of the a priori SNR to achieve fast noise tracking. Compared to three state-of-the-art approaches, i.e., minimum statistics (MS), MMSE-based approach, and speech presence probability (SPP)-based approach, it is clear from experimental results that the proposed algorithm exhibits more excellent noise tracking capability under various nonstationary noise environments and SNR conditions. When employed in a speech enhancement system, improved speech enhancement performances in terms of segmental SNR improvements (SSNR+) and perceptual evaluation of speech quality (PESQ) can be observed.     

L1 Fourier spectral methods for a class of generalized two-dimensional time fractional nonlinear anomalous diffusion equations

In this paper, L1 Fourier spectral methods are derived to obtain the numerical solutions for a class of generalized two-dimensional time-fractional nonlinear anomalous diffusion equations involving Caputo fractional derivative. Firstly, we establish the L1 Fourier Galerkin full discrete and L1 Fourier collocation schemes with Fourier spectral discretization in spatial direction and L1 difference method in temporal direction. Secondly, stability and convergence for both Galerkin and collocation approximations are proved. It is shown that the proposed methods are convergent with spectral accuracy in space and $(2?\alpha )$ order accuracy in time. For implementation, the equivalence between pseudospectral method and collocation method is discussed. Furthermore, a numerical algorithm of Fourier pseudospectral method is developed based on two-dimensional fast Fourier transform (FFT2) technique. Finally, numerical examples are provided to test the theoretical claims. As is shown in the numerical experiments, Fourier spectral methods are powerful enough with excellent efficiency and accuracy.     

Combining GA and iterative MVDR for DOA estimation in space-time CDMA systems

This paper deals with direction-of-arrival (DOA) estimation of minimum variance distortionless response (MVDR) approach based on iterative searching technique for space-time code-division multiple access (CDMA) systems. It has been shown that the iterative searching technique is more likely to converge to a local maximum, causing errors in DOA estimation. In conjunction with a genetic algorithm for selecting initial search angle, an efficient approach is presented to achieve the advantages of iterative DOA estimation with fast convergence and less computational load over existing conventional spectral searching MVDR estimator. Simulation results are provided for illustrating the effectiveness of the proposed approach.     

Piecewise Linear Segmentation of Digital Contours in O(N. Log(N)) Through a Technique Based on Effective Digital Curvature Estimation

A novel and effective approach to piecewise contour segmentation in terms of straight line segments is described. The reported approach is based on a recently developed technique for digital curvature estimation that relies extensively upon digital signal processing techniques. When combined with an energy-based curvature compensation strategy, also shown here, such a framework allows not only the fast and accurate determination of the curvature at each of the points of the original contour, but also provides an effective means for multiscale contour analysis through Gaussian lowpass filtering. Straight line segments can be straightforwardly and speedily derived from the obtained curvature diagrams simply by looking for maximum curvature points. Considering that all the processing takes place in terms of one-dimensional signals (the parametrized representation of the discrete contour in terms of its x and y coordinates), one-dimensional fast Fourier correspond to the major computational demand required by the proposed techniques, thus implying O(N.Log(N)). Application examples are provided that fully illustrate the potential of the proposed framework for fast and accurate piecewise linear segmentation. Discussion on the real-time aspects of the proposed methodology as well as the design of an effective parallel/pipelined architecture for its execution have also been included and discussed.     

混合CORDIC在分裂基FFT中的应用

提出了一个基于CORDIC的分裂基FFT/IFFT处理器来计算2048/4096/8192点DFT。蝶形处理器的算术单元和旋转因子产生器采用CORDIC算法实现,所有的控制信号在片内产生。相比于存储旋转因子所需的ROM,CORDIC旋转因子所用ROM尺寸更小。与传统的FFT实现相比功耗减少了25%。     

基于FFT和DTFT的指数衰减复正弦信号参数估计算法

针对加性高斯白噪声环境中指数衰减复正弦信号频率和衰减因子的估计问题,提出一种结合FFT和DTFT的离散三谱线插值算法。在利用FFT进行频率粗估计基础上,再利用FFT幅度最大谱线及其左右两侧任意相等位置处的DTFT辅助谱线进行频率和衰减因子精估计,并通过迭代提高算法性能。仿真结果表明,当辅助谱线距离幅度最大谱线位置的绝对值小于0.5时,信号实际频率位于FFT两条谱线之间任意位置,算法性能均接近克拉美罗下界,参数估计精度高于同类算法。     

Blind synchronization scheme using the conjugate characters of the OFDM BPSK-modulated symbol

Previously, Beek?s scheme for timing and frequency offset estimation in the OFDM system employs cyclic prefix (CP) has been proposed under the assumption of independent identified distributed (i.i.d.) OFDM symbols. Actually, the real data in the OFDM modulated symbol, transferred by the inverse fast Fourier transform (IFFT), has the characters of complex symmetry. With these characters, more information in the whole OFDM symbol could be used for the timing and frequency offset estimation. In this paper, two conjugate symmetry characters of the OFDM BPSK-modulated symbol are used to achieve blind timing estimation algorithm in the OFDM systems. One is symbol-based symmetry and the other is CP-based symmetry. With these two conjugate characters applied to the proposed algorithm, the timing of the OFDM BPSK-modulated symbol could be derived. Under an AWGN channel, based on the performance of lose symbol timing rate and estimator mean square error, the proposed algorithm is with a tremendous improvement compared with Beek?s estimation method. Under a multipath fading channel, the results show that performance including lose symbol timing rate and estimator MSE with the proposed algorithm is better than those algorithms with Beek?s estimation method. In practical OFDM applied system, the OFDM BPSK-modulated symbol could be used to replace the preamble or training sequences in the standard to obtain an accurate timing and frequency offset estimation and to avoid the data rate decreasing with the proposed algorithm.     

3D CAD model retrieval with perturbed Laplacian spectra

This paper presents a novel approach to the 3D CAD model retrieval, whereby the 3D models are treated and matched as undirected graphs. While there is much success made in the matching of graphs based on their spectral decomposition, most of these approaches consider smooth surfaces and are not suitable for CAD models because of their complex topology and singular structure. In the proposed approach, the models are simplified based on the piecewise flat properties of the surfaces first, and a perturbed Laplacian spectrum approach is then applied to characterize the shape. These spectral values are used as samples for spectral distribution estimation. The perturbed spectral distributions of different models are then compared by their KL-divergence for model retrieval. The proposed approach is tested with models from known 3D CAD database for verification.     

Bias-corrected realized variance under dependent microstructure noise

The aim of this study is to develop a bias-correction method for realized variance (RV) estimation, where the equilibrium price process is contaminated with market microstructure noise, such as bid-ask bounces and price-change discreteness. Although RV constitutes the simplest estimator of daily integrated variance, it remains strongly biased, and many estimators proposed in previous studies require prior knowledge about the dependence structure of microstructure noise to ensure unbiasedness and consistency. The dependence structure is unknown however, and needs to be estimated. A bias-correction method based on statistical inference from the general noise dependence structure is thus proposed. The results of Monte Carlo simulation indicate that the new approach is robust with respect to changes in the dependence of microstructure noise.     

Fast calculation of α-stable density functions based on off-line precomputations. Application to ML parameter estimation

This paper proposes a fast algorithm to compute arbitrary α-stable PDFs and CDFs. The procedure is based on off-line precomputations of α-stable values on a grid of points in the α–β parameter space, as well as in a set of abscissa points. The grid is built by defining two quality measures and assuring that minimum values for these measures are reached on those points. Results indicate that the method here proposed is substantially faster than directly evaluating the standard expressions proposed by Nolan [1]. The proposed algorithm makes it possible to efficiently use estimation methods based on the evaluation of the PDF, such as the Maximum Likelihood (ML) principle, which gives asymptotically unbiased and efficient estimates. To this end, an ML-based estimation algorithm is provided as well.     

Analysis of windowing/leakage effects in frequency response function measurements

This paper analyses the errors on the frequency response function measurement of a transfer function due to finite window effects (leakage). First an analysis of the rectangular and the Hanning window is made. It will be shown that the leakage error consists of two components: a transient error due to initial and end condition effects, and an interpolation error due to the combination of neighbouring spectral lines. Starting from these insights an extremely simple expression to calculate the leakage induced bias and variance errors is generated. Eventually, a new ‘default’ window is proposed with slightly better properties. This allows a reduction in measurement time by 25% if the leakage errors dominate the disturbing output noise.     

基于噪声倒谱阈值频谱估计的语音活动检测

针对低方差频谱估计的语音活动检测(VAD)中Welch频谱估计方法计算量大的问题,提出利用倒谱阈值方法估计VAD中的噪声功率谱.该方法在静音时期为噪声的倒谱设置阈值,利用快速傅里叶变换计算频谱,再更新VAD中的判决阈值.算法复杂度分析与仿真结果表明,该方法的检测性能与Welch方法相当,计算量降低约18%,同时降低整个...     

基于遗传算法的宽带目标波束空间DOA估计

提出一种基于遗传算法的宽带目标到达方位(DOA)估计新方法,该方法首先通过傅立叶变换方法设计频域恒定束宽转换矩阵,并使用该矩阵将阵元空间的数据转换到波束空间,同时,利用宽带波束空间遗传算法进行DOA估计。计算机仿真表明:在低信噪比条件下该算法相比于CSS类子空间算法、CSS类最大似然算法和阵元空间遗传算法有更高的分辨概率,更小的均方误差。     

Cluster-based dynamic variance adaptation for interconnecting speech enhancement pre-processor and speech recognizer

A conventional approach to noise robust speech recognition consists of employing a speech enhancement pre-processor prior to recognition. However, such a pre-processor usually introduces artifacts that limit recognition performance improvement. In this paper we discuss a framework for improving the interconnection between speech enhancement pre-processors and a recognizer. The framework relies on recent proposals for increasing robustness by replacing the point estimate of the enhanced features with a distribution with a dynamic (i.e. time varying) feature variance. We have recently proposed a model for the dynamic feature variance consisting of a dynamic feature variance root obtained from the pre-processor, which is multiplied by a weight representing the pre-processor uncertainty, and that uses adaptation data to optimize the pre-processor uncertainty weight. The formulation of the method is general and could be used with any speech enhancement pre-processor. However, we observed that in case of noise reduction based on spectral subtraction or related approaches, adaptation could fail because the proposed model is weak at representing well the actual dynamic feature variance. The dynamic feature variance changes according to the level of speech sound, which varies with the HMM states. Therefore, we propose improving the model by introducing HMM state dependency. We achieve this by using a cluster-based representation, i.e. the Gaussians of the acoustic model are grouped into clusters and a different pre-processor uncertainty weight is associated with each cluster. Experiments with various pre-processors and recognition tasks prove the generality of the proposed integration scheme and show that the proposed extension improves the performance with various speech enhancement pre-processors.