Reliability of X-band Dual-polarization Phased Array Radars Through Comparison with an S-band Dual-polarization Doppler Radar

Based on the observations of a squall line on 11 May 2020 and stratiform precipitation on 6 June 2020 from two X-band dual-polarization phased array weather radars(DP-PAWRs)and an S-band dual-polarization Doppler weather radar(CINRAD/SA-D),the data reliability of DP-PAWR and its ability to detect the fine structures of mesoscale weather systems were assessed.After location matching,the observations of DP-PAWR and CINRAD/SA-D were compared in terms of reflectivity(Z_(H)),radial velocity(V),differential reflectivity(Z_(DR)),and specific differential phase(K_(DP)).The results showed that:(1)DP-PAWR has better ability to detect mesoscale weather systems than CINRAD/SAD;the multi-elevation-angles scanning of the RHI mode enables DP-PAWR to obtain a wider detection range in the vertical direction.(2)DP-PAWR’s Z_(H)and V structures are acceptable,while its sensitivity is worse than that of CINRAD/SA-D.The Z H suffers from attenuation and the Z_(H)area distribution is distorted around strong rainfall regions.(3)DP-PAWR’s Z_(DR)is close to a normal distribution but slightly smaller than that of CINRAD/SA-D.The K_(DP)products of DP-PAWR have much higher sensitivity,showing a better indication of precipitation.(4)DP-PAWR is capable of revealing a detailed and complete structure of the evolution of the whole storm and the characteristics of particle phase variations during the process of triggering and enhancement of a small cell in the front of a squall line,as well as the merging of the cell with the squall line,which cannot be observed by CINRAD/SA-D.With its fast volume scan feature and dual-polarization detection capability,DP-PAWR shows great potential in further understanding the development and evolution mechanisms of meso-γ-scale and microscale weather systems.

Detection of Breaking Waves Using X-Band Pulse Radar in the Nearshore Region

An X-band pulsed Doppler microwave radar has been used to determine the characteristics of breaking waves. Field experiments were conducted at the Shuang-Si estuary in the north of Taiwan in the winter of 2005. Analyses on maxima radar cross section and Doppler frequency shift are done to characterize wave breaking zones. Based on observations of breaking waves, the wave breaking zones are shown to be located at water depths of 1.8 to 2.2 m in the experimental site. In general, the results indicate that a radar system has the potential to delineate the spatial variation of breaking waves clearly and that this is sufficient to achieve a measurement operation for near-shore air-sea interaction events.

复合微多普勒信号的时频谐波分析方法

在对复合微动目标的雷达回波进行时频分析时,存在时频曲线能量难以聚焦等问题,影响目标微多普勒特征提取和分析。对此,提出了时频谐波条纹的概念以及相应时频变换方法。基于复合微动包含中心运动分量和高频运动分量的假设,首先利用样条函数拟合中心分量频率曲线,构造时频变换核函数。然后,通过参数化时频变换将时域的复合微多普勒信号转化为时频平面上的谐波条纹。所提方法相比短时傅里叶变换和伪魏格纳分布等经典方法,能够明显改善复合微动目标雷达回波时频变换的能量聚集性与杂散抑制效果,仿真验证了所提方法的有效性和对低信噪比条件的适应性。

基于24 GHz连续波多普勒雷达能量密度分布的非接触式睡眠动作检测

睡眠动作是反映睡眠质量的重要生理指标。现有基于雷达的睡眠动作检测方法主要根据雷达信号原始能量的变化检测睡眠动作。由于不同目标动作幅度的不同会导致能量变化的差异,这些方法在检测不同目标的睡眠动作时准确率受限。为提高睡眠动作检测的准确率,提出了一种基于雷达能量密度分布的睡眠动作检测方法。首先,提出基于连续波多普勒雷达能量密度分布的检验统计量特征和显著性水平特征;随后基于上述特征引入XGBoost模型实现睡眠动作检测;最后,通过多种环境下的真实实验验证算法的有效性与鲁棒性。实验结果表明,所提出的检验统计量特征和显著性水平特征可以在降低训练样本需求的同时有效提高准确率,在多种环境下均能达到95%以上的检测准确率。

降水微物理过程的C波段垂直指向雷达反演研究

降水微物理过程是对空中雨滴的蒸发、碰并和凝结等演化过程的最直接描述。对降水过程中微物理特征的分析,有利于提高对云体的认识,为降水估计与数值模式参数化提供技术支撑。研究依托中国气象科学研究院灾害天气国家重点实验室在广东龙门建立的超级观测站,基于C波段调频连续波雷达高时空分辨率的多普勒功率谱数据。分析了2020年6月6-9日(前汛期),受西南季风影响,形成的一次持续时间长、累积雨量大、降水分布不均的降水过程中的几种典型对流单体的空中微物理特征。

基于微多普勒特征的人体动作识别

针对传统的视觉识别人体动作易受到光照变化、被遮挡和隐私问题等的影响,提出一种基于77 GHz毫米波雷达识别人体动作的方案。首先,使用77 GHz毫米波雷达采集人体动作的雷达回波样本,处理回波样本得到微多普勒频谱图;其次,首次提出一种新的自适应参数整流线性单元激活函数,将该激活函数与Resnet-18网络相结合;最后,将微多普勒频谱图数据集放入该网络训练并分类。实验结果表明,该方案对5种人体动作的平均识别准确率高达97.56%,较Resnet-PReLU网络提高了1.78%,有效地提高了对人体动作的识别精度。

Range Ambiguity Elimination in a Short-Range FMCW Radar System

Modified implementation architecture for sinusoidal frequency modulation is introduced to extract the range information from the received radar echo. Range ambiguity problem arises because the range is calculated from the estimated phase of the received signal which is wrapped into （0, 2π]. By integrating Doppler frequency shifts, the variation of range can be estimated and used as an auxiliary information to help eliminating the corresponding range ambiguity. The performance of the new technique is evaluated by simulations. The results show that this technique is robust to sever phase noise and can be used effectively for ambiguity elimination of the modified sinusoidal frequency modulated continuous wave radar.

Analysis on an Easterly Wave Thunderstorm and Gale Weather during the Latter Flood Season of 2010

[Objective] The research aimed to analyze an easterly wave thunderstorm and gale weather during the latter flood season of 2010.[Method] Based on conventional observation data,data of automatic station,NCEP 1°×1° reanalysis data,Doppler radar reflectivity and radial velocity data,the easterly wave thunderstorm and gale weather process which happened during 4-5 August,2010 was analyzed.The circulation situation,wind field and dynamic & thermal structures of easterly wave,echo characteristics of Doppler radar when thunderstorm and gale happened were studied.[Result] The thunderstorm and gale weather happened in the northeaster zone in front of the deep thick easterly wave trough.When thunderstorm and gale weather happened,position of the subtropical high was by north(ridge line was at 35° N).Eastward movement of the north branch of trough compelled subtropical high splitting into east and west circles.Westward extension of the east subtropical high and saddle-type field circulation were favorable for generation and development of the easterly wave.As development and westward movement of the easterly wave,in the influence zone in front of the trough,upper easterly component transmitted downward.Then,low-level northeaster obviously strengthened,and thunderstorm and gale weather appeared.In the zone where thunderstorm and gale happened,airflow had cyclonic convergence at the middle and low layers and anti-cyclonic divergence near the ground layer.Subsidence movement was obvious.Moreover,thermal force structure that it tended to be dry and cold at the middle and high layers,warm and wet at the low layer existed.Supercell storm was at the developed stage on 4th.It had obvious mesocyclone and hooked echo characteristics(meso-and small-scale characteristics).Characteristics of the squall line,gale region and adverse wind region appeared in central west Guangdong on 5th.Doppler characteristics of the mesocyclone,hooked echo,squall line,gale region and adverse wind region had indication significance for short-imminent forecast of the thunderstorm and gale.[Conclusion] This research revealed some information which had significance for forecasting local easterly wave thunderstorm and gale weather.

Iterative physical optics model for electromagnetic scattering and Doppler analysis

An iterative physical optics（IPO） model is proposed to solve extra large scale electric electromagnetic（EM） scattering from randomly rough surfaces. In order to accelerate the convergence of the IPO model, the forward-backward methodology and its modification with underrelaxation iteration are developed to simulate the rough surface scattering; the local iteration methodology and the fast far field approximation（Fa FFA） in the matrix-vector product are proposed to reduce greatly the computational complexity. These techniques make Monte Carlo simulations possible. Thus, the average Doppler spectra of backscattered signals obtained from the simulations are compared for different incident angles and sea states. In particular, the simulations show a broadening of the Doppler spectra for a more complicated sea state at a low grazing angle（LGA）.

基于无效目标滤除的毫米波雷达农田障碍物检测方法

农田障碍物检测是农机自动驾驶的前提,针对毫米波雷达的农田障碍物检测,本文提出一种无效目标滤除的算法。首先对毫米波雷达所输出的目标信息进行解析并提取农田目标障碍物信息,然后使用无效目标滤除算法对解析后的空目标、伪目标、非威胁数据进行滤除。对雷达数据中目标距离为零的空目标直接滤除;对因雷达工作性能或回波信号不稳定而产生的伪目标通过雷达有效目标生命周期法进行过滤;对超出横向距离阈值和纵向距离阈值的非威胁目标直接滤除。试验结果表明,所提算法在静止状态,平均滤除率达到85%以上。当速度为3 km/h时,未作业状态的平均滤除率为85.24%,旋耕为84.23%;当速度为5 km/h时,未作业为84.22%,旋耕为84.18%;当速度为7 km/h时,未作业为84.19%,旋耕为84.16%。尽管在行驶状态下,随着速度的提升或悬挂旋耕机,滤除率有一定的下降,但本文所提算法在各种状态下的平均滤除率均可达到84%以上。

基于多源数据的云中过冷水反演方法研究

云中过冷水的含量及分布是造成飞机积冰的主要原因,因此准确识别云中过冷水区域尤为重要.利用英国Chilbolton观测场的35GHz、94GHz毫米波测云雷达、激光雷达以及微波辐射计,结合探空资料,对阈值法和模糊逻辑算法识别云中过冷水分布情况进行对比研究,以及利用多普勒密度算法对不同水凝物含水量情况进行计算.结果表明:1)对比阈值法以及微波辐射计液态水路径的反演结果后认为模糊逻辑算法对水凝物的识别具有更高的反演精度;2)混合相态云中由于冰晶粒子主导雷达回波强度,会导致云中液态水含量被低估,需分别计算不同水凝物的雷达反射率因子.

面向毫米波雷达手势分类的RDI数据增强方法

在基于毫米波雷达的手势分类任务中,应用深度学习技术可以显著提高准确率。然而,深度学习技术对数据量的依赖性很高,当训练样本数据稀缺时容易出现过拟合问题。由于不同的毫米波雷达参数差异较大、采集数据耗时费力,基于毫米波雷达的手势数据量往往非常有限。为了解决数据量稀缺这一问题,本文提出了一种融入注意力模块的距离多普勒图自编码(Range-Doppler Image AutoEncoder with Attention Module,RDI-AEAM)数据增强方法,旨在增强毫米波雷达手势数据的RDI表征。该方法针对RDI缺乏语义信息、难以进行标注以及特征不明显的特点,构建了一个融入注意力模块的自编码网络。首先,利用自编码器进行特征提取和数据压缩,学习输入数据的分布并提取有用特征。其次,利用注意力模块专注学习通道和空间维度的特征,解决特征不明显问题,使模型能够更加集中关注重要特征。训练过程中,预定义了原始数据标签,使用最小均方误差损失函数衡量生成数据的质量,达到设定阈值时将生成数据与预定义标签相关联,而无须额外后期标注。实验先选择100%训练集进行增强,相比仅使用原始训练集进行训练的结果,数据增强后的准确率在自建数据集上提高了0.83%,在公开数据集deepSoli和VR-HGR上分别提高了0.39%和3.23%,表明RDI-AEAM方法提高了手势分类性能。本文进一步探究了采用更少原始数据的增强效果,使用25%的训练集进行增强,在三组数据集上分别取得1.92%、2.62%和1.56%的提升。

涡旋电磁波雷达平动旋转目标三维微动参数提取方法

与传统平面电磁波雷达相比,涡旋电磁波雷达能同时观测到目标投影到雷达径向和垂直于径向平面的微动分量,可为目标识别提供更多信息。当前关于涡旋电磁波雷达微多普勒效应的研究尚处于起步阶段,初步实现了对旋转目标的三维微动参数的提取,但均未考虑目标平动的影响。因此,该文研究了涡旋电磁波雷达中平动旋转目标的微多普勒效应,推导了平动旋转目标的角多普勒性质,提出了基于1/4微动周期多普勒频移曲线的三维微动参数提取方法,获得了目标旋转频率、旋转半径、旋转矢量和平动速度矢量等参数。仿真验证了角多普勒性质的正确性和参数提取方法的有效性。

基于多普勒域补偿的车载雷达距离角度联合成像算法

高性能、高分辨率单快拍前视成像技术是赋能车载雷达发展的关键,但距离/多普勒走动问题会限制相干积分的实施,同时系统分辨率也往往受限于硬件参数难以提高。根据车载毫米波雷达时分多输入多输出(TDMMIMO)的前视成像体制,该文提出多普勒域补偿和点对点回波校正方法,完成多域信号解耦合,同时完成距离多普勒走动校正和多普勒解模糊。由于有限阵元数及强噪声干扰限制了传统单维度距离角度成像准确性,因此,该文提出一种基于改进贝叶斯匹配追踪方法(IBMP)的多域联合估计算法。该方法基于伯努利-高斯(BG)模型,在最大后验(MAP)准则约束下迭代更新估计参数和支撑域,实现了多维联合信号的高精度重构。仿真和实测结果表明该文方法能够有效解决距离走动问题,并提高雷达前视成像的角度分辨率,具有较强噪声鲁棒性。

融合快照集成与知识迁移的人体行为识别算法

用卷积网络进行人体行为识别是毫米波雷达的一个热门研究方向。由于卷积网络结构的缺陷性,而且目前用于人体行为识别公开的雷达领域数据样本量过少,传统深度学习算法对雷达微多普勒图像的识别率不高,且在训练过程中容易出现过拟合的现象。针对上述问题,本文提出一种融合快照集成与迁移学习的雷达人体行为识别算法。首先,针对深度卷积网络无法提取图像全局特征的问题,该算法通过搭建Vision Transformer(VIT)模型引入注意力机制。其次,通过VIT模型在公开自然数据集上进行任务迁移和特征空间的迁移,解决微多普勒图像的识别过拟合的问题。最后,利用基于快照集成的投票机制算法,提升模型对复杂雷达微多普勒图像的识别能力。试验结果表明,在目标任务数据集样本量少、背景复杂的情况下,该算法能在不增加训练成本的前提下提升微多普勒图像的识别准确率,在VIT模型下该算法识别准确率达到了89.25%,优于经典卷积神经网络。

基于涡旋电磁波的雷达应用研究进展

携带轨道角动量的涡旋电磁波,相比于传统平面波有着独特的螺旋相位结构与模态正交特性,因而能获取更多信息,在雷达探测中受到广泛关注。轨道角动量调制带来的升维信息为新体制雷达成像提供了条件;另外,利用涡旋电磁波观测到的旋转多普勒效应,可获得平面波难以探测的垂直于距离向的运动信息,弥补了平面波雷达探测的缺陷。文中首先介绍了涡旋电磁波的基本概念、特点及常见的产生方法;然后,介绍了基于均匀圆阵的涡旋电磁波成像基本原理,简单推导了利用涡旋波进行旋转多普勒检测的基本原理,梳理了涡旋波在雷达成像和旋转多普勒检测两个应用领域的研究进展;最后,对未来的研究趋势进行了展望。

Convolution neural network and 77 GHz millimeter wave radar based intelligent liquid classification system

An intelligent liquid classification system based on 77 GHz millimeter wave radar and convolution neural network are proposed in this paper.The data are collected by the AWR1843 radar platform and processed by the neural network on the host PC in real-time.The doppler heatmap generated by radar signal processing is tried for the first time as the input of the system.The information carried by the heatmap in 2 dimensions is analyzed and the reason why the doppler heatmap could be used for classification is explained.The feasible experiment proved that the proposed method can successfully classify 8 kinds of common liquid with high accuracy.The result of the experiment is explained and the limitations of the experiment are discussed.It can be drawn that the combination of FMCW millimeter wave radar and convolution neural network is a method with great potential for liquid classification.The advantages of real time,non-invasive and unilateral measurement can also be used for the detection of dangerous liquids.

基于原型与主动学习的天波雷达干扰检测方法

本文研究天波雷达基于距离⁃多普勒(Range⁃Doppler,RD)图像的干扰检测问题。在干扰检测过程中,错误检测可能是干扰的漏检与虚警问题,为此考虑采用主动学习方法,将算法模型难以判决的样本由专家标注,并将标注样本加入至训练集中以达到提升检测性能的目的。同时,也需要解决训练集样本的冗余问题,为此使用原型数据学习方法,建立有干扰和无干扰样本数据云,有效地降低训练集样本量。实测数据实验表明,原型方法将初始训练集样本数量降低至23.5%,主动学习方法取得的检测准确率为97.42%,而传统监督学习最近邻方法准确率为87.96%。因此,本文方法能够有效提升天波雷达干扰检测能力,为天波雷达是否需要进行干扰处理与换频检测等工作提供可靠依据。

基于微多普勒信号分离和SqueezeNet的人体身份识别

针对基于雷达传感器的人体身份识别问题,本文提出一种基于微多普勒信号分离和SqueezeNet的人体身份识别方法。首先利用雷达对人体行走的步态进行探测并收集回波数据,回波数据经过预处理得到微多普勒时频谱图;其次用阈值法对时频谱图进行微多普勒信号分离从而得到四肢的时频谱图;最后将其输入到SqueezeNet网络,采用Softmax分类器来实现人体身份识别。实验结果表明,经过微多普勒信号分离后人体身份识别准确率提高5.25%,SqueezeNet网络与其他网络相比,在网络性能上具有准确率高、网络参数少、测试时间短等优势。

高频地波雷达东海海洋表面矢量流探测试验

OSMAR2 0 0 0高频地波雷达系统采用线性调频中断连续波体制 ,天线阵收发共用 ,应用多重信号分类法(MUSIC)和最小方差法 (MVM)获得海面回波到达角 .2 0 0 0年 10月 2 3～ 2 9日进行了OSMAR2 0 0 0同传统海洋测量设备以及其他雷达系统的对比验证试验 .文中首先介绍雷达系统和提取海流信息所用的回波到达角估计算法 ,然后给出这次试验中海洋表面矢量流的有关对比测量结果 .结果表明了所研制雷达系统以及所采用信号处理方法在海洋表面流探测中的有效性 .