一种低信噪比环境下的语音端点检测算法

端点检测技术是语音信号处理的关键技术之一，为提高低信噪比环境下端点检测的准确率和稳健性，提出了一种非平稳噪声抑制和调制域谱减结合功率归一化倒谱距离的端点检测算法。该算法首先通过抑制非平稳噪声再采用调制域谱减消除残余噪声来提升信噪比，减少语音失真。然后再提取每帧信号的功率归一化倒谱系数，计算每帧信号与背景噪声的功率归一化倒谱距离。最后将该倒谱距离作为检测参数，采用双门限判决方法进行端点检测。实验结果表明，该端点检测算法对语音帧和噪声帧具有较好的区分性。此外，在低信噪比环境下，所提出的算法对于不同类型的噪声都具有较好的稳健性。     

Very sensitive measurement method of electron devices current noise

The problem of measuring very low levels of current noise in bipoles (linear or not) is dealt with, and a measurement technique is proposed. This technique allows the measurement of noise power spectra 6-10 dB lower than the equivalent input power spectrum of the amplified necessary to perform the measurement. An improvement of 16-20 dB in the sensitivity is obtained with respect to the one of conventional methods, which, for an acceptable accuracy, require the noise of the bipole under test to be 10 dB larger than the equivalent input one of the amplifier. The present method is based on the accurate measurement of the amplifier transimpedance with respect to the input current noise sources and on the precise evaluation and subtraction of the contribution from all the spurious sources to the total noise. The whole procedure is implemented by means of a dual-channel signal analyzer and almost completely automated. The technique has been tested by using it to measure the power spectra of the noise given by known generators, of the Nyquist noise produced by bipoles made up of resistors and capacitors, and of shot noise in p-n junctions. The experimental results agree very well with theoretical predictions     

Cutoff frequencies and noise power law model of spectral density:adaptation of the multivariance method for irregularly spaced timingdata using the lowest-mode estimator approach

The concept of structure functions, which is an extension of the variance approach, is useful to determine the variance (the structure function) which is optimized for a type of noise and for an order of drift. The multivariance method was developed to use different variances over the same signal. It is then possible to select a set of variances in which each variance is optimized to the determination of one parameter (of one noise level, drift, or cutoff frequency). Recently, we adapted this method to irregularly spaced timing data. In this connection, we replaced the structure functions by another method of spectral density estimation: the lowest-mode estimator, introduced by J.E. Deeter and P.E. Boynton (1982, 1984) for the analysis of pulsar timing data. Different lowest-mode estimators can be constructed according to two priorities: the order of drifts that must be removed and the type of noise for which the sensitivity must be maximum. Thus, a multivariance system is developed using a set of different estimators. The details of this method are described, and the results for different signals are discussed in this paper     

Research on ultrafine particle size measurement using small amounts of data

The paper puts forward a new method of ultrafine particle size measurement using small amounts of data of a dynamic light-scattering signal, and establishes an arithmetic model of the measurement by wavelet package transform. First, through the wavelet package transform, the ultrafine particle dynamic light-scattering signals were decomposed into multifrequency bands. Then, the noise of signals of different frequency bands were removed and the power spectrum of the wavelet packet coefficients of each frequency band was calculated. Finally, the ultrafine particle size distribution information could be deduced from inversing the power spectrum. The standard polystyrene particles of 100, 300, and 400 nm were measured using this method, and the inversion results indicated that this method can effectively remove noise and improve the accuracy of particle size measurement using small amounts of data.     

Characterization of Frequency Stability: A Transfer Function Approach and Its Application to Measurements via Filtering of Phase Noise

Frequency stability of high-quality signal sources is characterized in the Fourier frequency domain by the spectral density Sy(f) of the fractional instantaneous frequency deviation y(t), and in the time domain by the Allan variance ?y2(?). Two well-known types of measuring apparatus used to evaluate these parameters are analog spectrum analyzers and digital electronic counters, respectively. A detailed analysis of the structure of the relation between ?y2(?) and Sy(f) shows that it is possible to define a variance, i.e., a time-domain measure, by its transfer function in the Fourier frequency domain, even when no corresponding measurement sequence exists in the time domain. Two different kinds of variance are then defined, which possess different properties for white and flicker phase noises. One of these variances is an estimate of the Allan variance. These variances may be measured by a suitable filtering of phase noise at the output of a phase detector.     

Some problems related to the quantitative evaluation of physical properties of the porous medium from flow tests with non-newtonian fluids

This paper analyses two methods for determining the coefficients appearing in modified Darcy's law describing the flow through a porous medium in which fluids are non-Newtonian governed by the power law with a yield stress. The first method uses the measurement data in flow tests of steady state, while the second is based on the pressure response measurement during the transient flow of short duration, recorded at a fixed location. An analysis of non-Newtonian flow behaviour associated directly with quantitative interpretation of the data measured from these techniques is presented and discussed. It has been shown that both techniques have advantages and disadvantages and selection of one of them should experimentally be solved in terms of accuracy degree required in obtaining the coefficients in modified Darcy's law for non-Newtonian fluids.     

An Evaluation of the "Back-to-Back" Method of Measuring Adaptor Efficiency

The problem of adaptor evaluation is of considerable interest in the UHF and microwave art where, because of the several different types of transmission line in common use, there is a frequent need to extend an established measurement capability in one type of line to other types of lines, e.g., from rectangular waveguide to coax. In power and noise calibrations, it is the adaptor efficiency which is usually of interest, and one indication of the adaptor losses may be obtained by connecting an identical pair "back to back" and measuring the insertion loss of the combination by the usual techniques. If, then, the losses are assumed to divide equally between the two, the efficiency is thereby determined. As a practical matter, the losses do not, in general, divide equally, which leads to an error whose evaluation is the subject of this paper. In particular it is shown that the method is capable of good accuracy provided that the assumed identity of the two adaptors is satisfied.     

Study of the excess noise associated with demodulation of ultra-short infrared pulses

The demodulation of ultra-short light pulses with photodetectors is accompanied by excess phase noise at the pulse repetition rate and harmonics in the spectrum of the photocurrent. The major contribution to this noise is power fluctuations of the detected pulse train that, if not compensated for, can seriously limit the stability of frequency transfer from optical to microwave domain. By making use of an infrared femtosecond laser, we measured the spectral density of the excess phase noise, as well as power-to-phase conversion for different types of InGaAs photodetectors. Noise measurements were performed with a novel type of dual-channel readout system using a fiber coupled beam splitter. Strong suppression of the excess phase noise was observed in both channels of the measurement system when the average power of the femtosecond pulse train was stabilized. The results of this study are important for the development of low-noise microwave sources derived from optical "clocks" and optical frequency synthesis.     

基于表面振动法的柴油机辐射噪声测量和分析

以4120SG柴油机为研究对象，提出基于激光测振的表面振动法测量柴油机辐射噪声。重点讨论了结构辐射系数的确定及其影响因素，提出了不同结构部件计算辐射系数的方法。利用表面振动法计算了柴油机主要部件的声功率，识别了主要噪声源。通过与噪声测量结果的对比，验证了方法的准确性。     

一种微波源调频噪声测量系统

分析了一种微波源调频噪声测量方法的原理，根据需要研制了一套微波源调频噪声测量系统，详细给出了为保证系统测量结果准确可靠所必需的调试步骤，并结合实际设计出一种独特的定标方法，使用该系统实测了一个２２．６８ＧＨｚ锁相稳频信号的调频噪声，并对测量结果进行了讨论。     

基于机器视觉的丝网印刷样板尺寸测量方法

丝网印刷样板种类和样式多,每片样板上待测目标也较多,不宜采用一致的梯度阈值参数定位边缘实施测量;此外,传统视觉检测采用定位矩形框建立测量坐标系的方式会引入测量误差,因此,提出一种基于机器视觉的丝网印刷样板的尺寸测量方法。采用一种由粗到细的测量策略,借助每一类待测目标的模板进行信息统计,针对性地设置阈值参数,改善测量精度。在测量阶段,利用基于图像金字塔和归一化互相关函数相结合的分层匹配算法实现多个待测目标的粗糙定位,再使用由模板信息统计获取的阈值参数进行边缘精细定位,建立局部坐标系,完成测量。实验结果表明:在同等实验条件下,此方法能够有效提高测量精度,算法改进后测量结果的平均相对误差由原来的4.02%降到1.47%,并且无需用户介入调整测量参数,适用于不同种类丝网印刷样板的灵活测量。     

Visualisation and analysis of noise sources of a gear unit

In technical acoustics, noise source visualization is an important tool. There are many different techniques of noise source visualization, most of which have been developed, bearing a specific noise source in specific type of acoustic environment in mind. Consequently, an individual visualization method is restricted to certain types of noise sources in a specific acoustic environment and in a limited frequency area. This article presents a new visualization method of complex noise sources using an acoustic camera. In this respect, a new algorithm is used, enabling visualization of all types of different complex noise sources. Monopole, dipole or quadruple noise sources can be observed simultaneously. A moving noise source can be tracked by means of an acoustic camera. In addition to that, various transient acoustical phenomena can be noted. On the basis of this, advanced determination of the condition of mechanical systems by means of noise is possible.     

浅海目标辐射噪声测量中声聚焦现象引起的误差分析及修正

研究了浅海近距离声聚焦特性及其对目标辐射噪声测量的影响和修正方法。利用基于射线理论的虚源法,建立了典型声源和复合声源的声场计算模型,通过具体算例分析了声聚焦引起的测量误差,将单点测量模型和阵列多点测量模型进行对比,并进行湖上试验验证,结果表明阵列多点测量模型能有效的对信道空间传递函数平滑,为如何减小声聚焦对目标辐射噪声测量的影响提供了依据,有助于实际工程测量的数据分析和测量精度的提高。     

火电厂主蒸汽流量测量方法研究

说明准确测量主蒸汽流量的重要性,介绍当前电厂采用的主要测量方法及其原理和特点,对比直接测量法的误差大小,对间接测量方法的原理进行说明,并提出了改进建议。     

Multivariate calibration of non-replicated measurements for the factored noise model

The accuracy of a multivariate calibration (MVC) model for relating concentrations of multicomponent mixtures to their spectral measurements depends on effective handling of errors in the measured data. For the case when error variances vary along only one mode (either along mixtures or along wavelengths), a method to estimate the error variances simultaneously along with the spectral subspace was developed by Narasimhan and Shah (Control Engineering Practice, 16, (2008), 146–155). This method was exploited by Bhatt et al. (Chemom. Intell. Lab. Syst., 85, (2007), 70–81) to develop an iterative principal component regression (IPCR) MVC model, which was shown to be more accurate than models developed using PCR. In this work, the IPCR method is extended to deal with measurement errors whose variances vary along both modes, by using a factored noise model. As a first step, an iterative procedure is developed to estimate the error variance factors along with the spectral subspace, which is subsequently used in developing the regression model. Using simulated and experimental data, it is shown that the quality of the MVC model developed using the proposed method is better than that obtained using PCR, and is as good as the model obtained using Maximum Likelihood PCR, which requires knowledge of the error variances. For dealing with large data sets, a sub-optimal approach is also proposed for estimating the large number of error variances.     

Random Errors in Time Interval Measurement Based on SAW Filter Excitation

This paper deals with the time interval measurement method that makes use of a surface acoustic wave (SAW) filter as a time interpolator. The method is based on the fact that a transversal SAW filter excited by a short pulse can generate a finite signal with highly suppressed spectra outside a narrow frequency band. If the responses to two excitations are sampled at clock ticks, they can be precisely reconstructed from a finite number of samples and then compared to determine the time interval between the two excitations. We have analyzed noise contributions to the overall uncertainty of the interval measurement. The following four noise sources have been considered: (1) noise of excitation; (2) noise of amplifier; (3) quantization noise of the analog-to-digital converter; and (4) clock and sampling jitter. The main goal of the analysis was to determine the uncertainty of the measurement in dependence on the variances of contributing noises, and the magnitude of the frequency response of the filter. First, each uncertainty component has been analyzed under general assumptions. Second, a general narrow-band filter has been investigated. Finally, we have found the solution for an ideal narrow-band filter where we have obtained simple illustrative results. The results have been compared to Monte-Carlo simulations. Good agreement between theory and simulation has been demonstrated except for the outliers due to the false ambiguity solution that occurs with large measurement errors and small relative filter bandwidth.     

三维空间有源消声的一种通用优化方法

本文根据声源辐射理论、声场叠加原理以及测得的声功率大小与所选取的测量面无关这一特点,提出了一种通过最小化初级声源和次级声源总的声功率对三维空间噪声的主动控制进行优化的数值方法。文中采用边界元近似,将总的声功率表示成次级声源复强度的二次型正实函数。本优化方法是Bullmore 等人提出的解析方法的推广,具有广泛的适用性,可用于设计初级声源在各面元上的声压可知、次级声源到各面元上的声传播可确定的任何场合下控制器优化传递函数。计算过程中仅用了几次复矩阵乘法和一组复系数线性方程组的求解,因此计算简便、速度较快。文中采用这种方法对圆柱壳和无幕活塞辐射器这两种具有典型指向特性的分布初级声源辐射噪声的优化控制问题进行了研究,说明了方法的有效性。文中还给出了实验验证结果。     

声学风洞中的高速列车模型气动噪声试验研究

研究借助气动-声学风洞试验平台,首先针对某高速列车的1：8缩尺比例的三车编组模型建立了气动噪声试验方法和突显不同的噪声源的模型处理方法,并结合流场外自由场传声器和传声器阵列的测量结果,分析了模型上的主要噪声源特性及对整个模型的贡献量大小。研究表明：转向架和受电弓噪声是模型的最主要噪声源,其次是车连接部位间隙,再次是鼻尖和排障器,最后是尾车,同时,并给出了这些噪声源的特性,这对于认识高速列车气动噪声和改善设计有重要的参考价值。研究也说明所提出的试验研究方法是一种研究高速列车气动噪声较为有效地方法。     

Interactions of low-power photons with natural opals—PBG materials,photonic control,natural metamaterials,spontaneous laser emissions,and band-gap boundary responses

One overall goal of this research was to examine types of naturally-occurring opals that exhibit photonic control to learn about previously-unknown properties of naturally occurring photonic control that may be developed for broader applications. Three different photon sources were applied consecutively to three different types of natural, flawless, gem-quality precious opals. Two photon sources were lasers (green and red) and one was simulated daylight tungsten white. As each type of precious opal was exposed to each of the photon sources, the respective refractions, reflections, and transmissions were studied. This research is the first to show that applying various pleochroic and laser photon sources to these types of opals revealed significant information regarding naturally occurring photonic control, metamaterials, spontaneous laser emissions, and microspheroid cluster (inter-PBG zone) boundary effects. Plus, minimizing ambient light and the use of low power photon sources were critical to observing the properties regarding this photonic materials research. This research yielded information applicable to the development of materials to advance applications and devices of photonics, phononics, optoelectronics, nanomaterials, and metamaterials.     

基于深度学习的舰船辐射噪声多特征融合分类

为了改善分类系统的性能,进一步提高舰船辐射噪声分类的正确率,该文提出了一种基于深度神经网络的多特征融合分类方法。该方法首先提取舰船辐射噪声几种不同的特征,将提取的特征同时用于训练具有多个输入分支的深度神经网络,使网络直接在多种特征参数上进行联合学习,通过神经网络的输入分支和连接层实现特征融合,再对舰船辐射噪声进行分类。为了特征深度学习提取了舰船辐射噪声的频谱特征、梅尔倒谱系数和功率谱特征,并将多特征融合分类方法与在一种特征上进行深度学习分类方法的正确率进行对比。实验结果表明,基于深度学习的多特征融合分类方法可以有效地提高舰船辐射噪声分类的正确率,是一种可行的分类方法。