机械振动信号盲源分离的时域方法

介绍盲源信号分离基本原理,分析了瞬态混叠模型和卷积混叠模型的特征,给出两个源信号独立性的评判准则,研究了盲分离正确性的评价误差函数,进行数值仿真,并用时域卷积瞬态混叠模型对实际机械振动信号进行源信号的盲分离。数值仿真和实际信号的分离效果较好,说明所用方法在机械振动信号盲源分离中的适用性。文中还分析了多源信号分离中所用方法的收敛性。     

基于参考信号的广义特征值分解盲源分离算法及其在北斗多路径效应削弱中的应用

多路径误差是北斗导航定位系统高精度动态监测的主要误差源。针对北斗导航定位系统多路径误差的特性,结合广义特征值盲源分离方法的优势,提出一种基于参考信号的广义特征值盲源分离算法来削弱多路径效应的影响。首先将前一天的原始坐标残差序列通过奇异谱分析方法进行去噪,其结果作为初始参考信号;然后将当天的原始坐标残差序列进行经验模式分解方法分解,分解得到的IMF分量作为虚拟观测数据,利用广义特征值盲源分离算法获取当天多路径误差信号;最后,利用仿真数据和连续10天的实际观测数据进行试验分析,结果表明利用该方法建立的多路径误差改正模型能有效地了削弱多路径的影响,北、东、天三个方向精度分别提高了78.8%、35.3%、90.1%。提出的模型在一定程度上解决了固定多路径模型随着时间推移重复性减小且有效性降低的问题。     

基于Lamb波频散特性的薄板声发射源定位方法研究

声发射源时差定位方法中波速的确定是定位准确的关键问题,根据模态声发射理论,声发射信号在薄板内的传播过程中具有频散现象和多模态特性.不同频率不同模态的声波其传播速度不同.基于Lamb波频散特性和声发射检测技术,采取时频联合分析方法对铝薄板中声波模式进行识别;利用低频段频散不明显的扩展波的波速和同一频率同一模态波到达两传感器的时间差来实现声发射源的准确定位;通过铝薄板中AE源定位实验表明,采用S0模式,即扩展波的波速进行定位计算,可以比较准确地确定声发射源的位置,而且理论值与实际值相差很小.     

基于蜂窝网格粒子滤波的行人导航航向估计方法

为了提高低成本行人惯性导航系统的实时导航精度,针对行人惯性导航系统航向角发散问题,提出一种适用于实时行人导航系统的蜂窝网格粒子滤波算法(CSPF,cellular structure particle filter)。该方法将二维平面划分为紧密相连的蜂窝六边形区域,记录行人运动的历史轨迹信息,采用粒子滤波方法对航向误差补偿量进行估计。蜂窝网格粒子滤波算法在行人惯性导航系统的实验结果表明,引入了蜂窝网格粒子滤波的行人惯性导航方法与常规方法相比能有效改善实时行人惯性导航精度。在二维行走实验中,水平位置误差百分比由5.3%改进到1.7%;三维行走实验中,水平误差百分比由4.2%改进到2.9%。     

基于行为概率分析的室内行人导航方法

行人在室内场景下进行混合步态运动时,传统零速检测方法易出现“漏检”和“误检”,造成定位精度下降。针对此问题,提出了一种基于行为概率分析的行人零速检测与爬楼高度估计方法。一方面,设计了基于步态周期检验的最大概率零速点提取方法,通过足部零速广义似然比概率统计对步态周期进行准确划分,继而实现对步态周期内最大概率零速点的提取;另一方面,采用基于长短期记忆的深度学习网络模型实现上下楼运动检测,基于气压高度计与惯性器件数据,构建描述足部跨越台阶数的能量概率函数,并对台阶高度、跨越台阶数以及爬楼高度进行实时在线估计。最后开展了行人室内导航实验,相较于传统零速检测算法和零速阈值自适应调整算法,所提算法的平均零速点提取成功率分别提高24.44%和16.21%,平均导航定位性能分别提高了55.14%和39.15%,高度误差由7.79%降低到0.39%。     

基于MEMS惯性测量单元的多源信息自适应步数检测方法

针对基于MEMS惯性测量单元的行人航迹推算中步数检测方法仅利用单一的加速度信号检测精度较低的问题,提出一种多源信息自适应步数检测方法。该方法通过综合考虑人体运动过程中的加速度信号和角速度信号,根据不同的步态特征通过设定不同的自适应阈值条件实现步数的检测。虽然常规的峰值检测算法和固定阈值检测算法在单一步态下步数检测精度相对较高,但是对复杂运动状态下的步数检测精度很差,无法适用于真实的行人运动过程中步数的检测。然而多源信息自适应步数检测方法却能够在行人不同运动状态下精确检测步数,该方法明显优于常规的峰值检测方法和阈值检测方法。试验结果表明,本文提出的多源信息自适应阈值检测方法在行人不同运动状态下的步数检测精度可达98%以上。     

缺陷源摆动的对传波

基于流体力学方程组,对长高比Γ=30腔体内混合流体对流中缺陷源摆动的对传波的特性进行了数值模拟.结果发现,当分离比Ψ=-0.6时,缺陷源摆动的对传波的缺陷源由两个对流圈同时被拉长同时被分裂后形成.对传波的存在区间为Dr=0.198,对传波的摆动周期较小并且基本都稳定在T_o=4.8.对传波的摆动幅度较小并且几乎不随着相对瑞利数r变化.当Ψ=-0.2时,对传波的缺陷源下方対流圈轴线与缺陷源移动方向保持一致,没有对传波分支产生.对传波的缺陷源上方存在多个与缺陷源移动方向几乎垂直的対流圈的轴线,存在对传波分支,在对传波分支上出现间歇性的缺陷.从而形成具有单侧缺陷的缺陷源摆动的对传波.对传波的存在区间为Dr=0.16.对传波的摆动周期为To=65.9-85并且随着相对瑞利数的增加而增加.对传波的摆动幅度较大,并且随着相对瑞利数的增加而减小.     

标准霍普金森压杆配置下的应力波分离及无时限实验数据处理

在经典一维应力波理论基础上以及试件受力平衡假定成立的条件下,提出了一种在标准霍普金森压杆实验配置下实现杆中左、右行应力波分离的新方法,可简单有效地解决常规霍普金森压杆在长时实验时左、右行波信号重叠的问题,从而保证实验中的全部应变测试数据都可以加以利用,显著提高了霍普金森压杆的测试能力。给出了新的基于杆中左、右行应力波信号的实验数据处理公式。作为霍普金森压杆实验中经典数据处理公式的扩展,在测试信号不需要进行波分离处理的情况下,新的数据处理公式等同于经典公式。利用ABAQUS有限元软件对霍普金森压杆实验进行了数值模拟,采用虚拟实验的方式,利用模拟测试点的应变信号进行了多种实验条件下的数据处理,对该应力波分离方法的有效性及误差进行了验证与评价。数值模拟结果表明,该应力波分离方法可以给出很好的数据处理结果。在标准霍普金森压杆上进行了部分实验并利用新的波分离方法及公式对数据进行处理,所得结果令人满意。     

基于多源信息融合的行人航位推算室内定位方法

传统行人航位推算(Pedestrian Dead Reckoning,PDR)定位技术存在严重的误差累积问题。针对因航向偏差引起的误差累积,提出一种借助建筑几何信息实现行人航向的实时补偿方案,通过提高定向精度来抑制定位误差的累积传递。分析利用外源绝对位置改善PDR定位结果,试验一种自适应模型噪声的扩展卡尔曼滤波(Extended Kalman Filter,EKF)滤波算法,实现PDR与WIFI定位源的滤波融合。通过实验对比分析,基于改正航向的PDR相较于传统PDR,有效抑制了误差的累积,将整体误差控制在5 m左右;传统PDR与WIFI源滤波融合,比单纯传统PDR提高了82.8%的精确度;航向改正PDR与WIFI源相融合,则比单纯传统PDR和航向改正PDR分别提高了90.2%和49.5%的精确度。结果表明:补偿改正航向和借助外源绝对位置滤波融合均可有效控制传统PDR的误差累积,根据条件约束可知航向改正PDR及其与WIFI源融合方案较适用于规则室内环境,而原始航向PDR与WIFI源融合方案则不受室内结构影响,在多次滤波后逐渐提高行人定位精度,从而可满足行人室内定位精度需求。     

一种基于双IMU框架的室内个人导航方法（英文）

为了实现低成本的室内行人导航,提出了一种双惯性测量单元(IMU)框架。在这种模式下,一个IMU固定于足部,另一个IMU固定于肩部。当行人在行走过程中处于静止状态时,卡尔曼滤波器利用测量得到的速度和角速度误差对足部IMU的解算误差进行预估,与此同时,通过对足部IMU和肩部IMU测量得到的航向角做差完成对航向角误差的观测。在此基础上,双IMU框架结构采用了闭环模式。实验结果显示,采用该方法能够提供行人导航信息,平均位置误差与采用开环模式的方法相比降低了14.93%左右。     

Investigation of the acceleration of aluminum particles behind a shock wave using instantaneous Laser Doppler Velocimetry

The acceleration of aluminum particles with a 5μm diameter in the flow field behind an incident shock wave was investigated experimentally in a 10-m long and 70 mm inner diameter shock tube. By means of instantaneous Laser Doppler Velocimetry (LDV) the velocity of the particles was observed directly. The light scattered by the moving particles is Doppler shifted and sent to the laser Doppler velocimeter. The velocimeter essentially consists of a phase-stabilized Michelson interferometer used as a sensitive spectrometer. An electro-optical circuit ensures the phase stabilization that results in a voltage signal independent of the scattered light intensity and proportional to the mean velocity of the particles at the measurement point. Because of the very short response time (1μs) of the LDV system used here, the latter gives a continuous real-time signal of the particle acceleration. To avoid particle oxidation the particles were accelerated by a high-speed nitrogen gas flow. From the measured velocity the dimensionless drag coefficient was calculated. The drag coefficient is related to the fluid dynamic force exerted by the gas on the particles. The experimental data were compared to theoretical models from the literature. A significant deviation between the model and the experimental data was observed. This deviation is supposed to be induced by the shock wave, which hits the particles and breaks them into pieces of a smaller diameter. Further experiments will be carried out in the future to check the size distribution of the particles after the shock has gone past them.      

Extraction of bulk wave characteristics from a pulsed ultrasonic polar scan

The pulsed ultrasonic polar scan (P-UPS) technique provides a map with characteristic contours reflecting the critical bulk wave angles, which directly relate to the in-plane elastic properties of the insonified material spot. Besides these contours, additional information is captured in a P-UPS experiment which has particular importance for the inspection of anisotropic materials. By extensive post-processing on a single P-UPS experiment, we successfully extracted the local directional (i) critical bulk wave angles, (ii) phase velocity profiles, (iii) slowness curves, (iv) walk-off angles and (v) energy velocity profiles, for the three different polarization states of bulk waves. The procedure is demonstrated and discussed for an isotropic aluminum sample as well as an autoclave manufactured orthotropic _[0]8${\left[0\right]}_8$ carbon/epoxy laminate. In addition, a numerical and experimental investigation of the role of the immersion liquid as a lens for the P-UPS result is performed. This lensing effect permits to zoom in/out on the P-UPS fingerprint, hence broadening the class of materials which can be inspected by the P-UPS technique.     

Explosive acceleration of liquid from a cylindrical container using the implosive technique

Tthe acceleration of a liquid from a cylindrical container by the products of instantaneous detonation of a superimposed explosive charge (implosion method) is studies numerically in a two-dimensional formulation. Analytical formulas for the asymptotic velocities of liquid expansion and acceleration of the container body were obtained using numerical calculations in a one-dimensional approximation. The effect of the fracture of the body on the velocity of liquid expansion was studied in a two-dimensional formulation.     

Reflection of disturbances caused by a weak energy-release source from a flat shock wave

The process of reflection of linear disturbances from a plane shock wave is considered in the case when these disturbances are caused by a weak energy-release source in a uniform supersonic flow of an inviscid non-heat-conducting gas. It is shown that, in the basic range of constitutive parameters, this interaction proceeds in such a way that the quantity characterizing the disturbance which provides a force load on the lateral surface of a body substantially changes when the reflection from the shock wave occurs.     

Reflection of disturbances produced by a three-dimensional heat source from a plane shock wave

The flow in the shock layer between a plane wedge and a plane attached shock is considered. The flow two-dimensionality is violated by the presence of a weak heat source in the shock layer. The relation between the mechanical actions of the disturbances singly reflected from the shock and those proceeding from the source and directly incident on the wedge surface is analyzed. The three-dimensional features of the field of the disturbances reflected from the shock are illustrated with reference to several examples.     

气相爆轰波正反射激波加速研究

实验采用压力传感器测量了指定点压力时间曲线。数值模拟基于二维反应欧拉方程和基元反应模型,采用二阶附加半隐的龙格-库塔法和5阶WENO格式分别离散时间和空间导数项,获得了指定点数值压力时间曲线。理论分析基于爆轰理论和激波动力学,分析了气相爆轰波反射过程所涉及的复杂波系演变并获得了反射激波速度。结果表明：本文数值模拟和理论计算定性上重复并解释了实验现象。气相爆轰波在右壁面反射后,右行稀疏波加速反射激波。其加速原因是：尽管激波波前声速减小,但激波马赫数增大,波前气流速度减小。在低初压下,可能还由于爆轰波后未反应或部分反应气体的作用,导致反射激波加速幅度比高初压下大。     

Oscillations of a shock wave induced by boundary layer separation

The causes of the oscillation of the separation-induced shock formed in supersonic flow over a step are investigated on the basis of measurements of the spatial correlations of the boundary pressure fluctuations.Translated from Izvestiya Akademii Nauk SSSR, Mekhanika Zhidkosti i Gaza, No. 4, pp. 168–170, July–August, 1990.