基于Mask R-CNN的电力关键设备运行状态检测

为判断电力关键设备两端负载电压与干路总电压之间的数值关系,提出基于Mask R-CNN的电力关键设备运行状态检测方法。在Mask R-CNN网络结构中,判断电量信号的时域范围与频域范围。然后根据负荷阻抗特征计算交流参数的取值结果,再联合已获取的电信号参量,求解连续相关函数,从而检测电力关键设备运行状态。实验过程中,设备两端负载电压、内阻消耗电压之和与干路总电压之间的差值未超过1.5 V,说明该方法能够证明电力关键设备两端负载电压、内阻消耗电压之和等于干路总电压数值的猜想成立,可以根据该验证检测电力关键设备运行状态是否正常。

改进Mask RCNN的盾构隧道渗漏水检测方法

渗漏水是盾构隧道结构存在潜在损伤或缺陷的重要表征,快速、准确检测出渗漏水位置,对隧道安全运营和维护具有重要意义。现有的方法大多采用光学影像对隧道渗漏水进行检测,受隧道内空间和光线条件限制,难以获得高质量病害图片。因此,本文提出了一种基于激光点云数据与改进Mask RCNN相结合的渗漏水检测方法。首先对激光点云反射强度进行修正;然后生成灰度图像并建立渗漏水病害数据集;最后在Mask RCNN算法中引入空洞卷积和变形卷积,实现了隧道渗漏水病害的快速检测。利用某地铁采集的数据进行验证,结果表明,本文提出的改进Mask RCNN算法相较于原始算法和FCN算法检测精度均有明显提升,在盾构隧道渗漏水识别方面性能表现较好。

基于Mask R-CNN的地质雷达岩溶预报图像识别研究

岩溶隧道开挖可能遭遇岩溶涌水、突泥等岩溶地质灾害,地质雷达能够有效预报岩溶等地质灾害。然而,传统地质雷达图像解译存在专家经验依赖性强、解译效率慢且易误判漏判等情况。本文采用可实现端到端识别的深度学习技术开展地质雷达图像目标检测与识别的研究,将基于Mask R-CNN的卷积神经网络算法应用于地质雷达岩溶预报图像异常的智能识别。在TensorFlow和Keras框架下,利用地质雷达设备采集获得的数据构建训练数据集和测试数据集,对Mask R-CNN深度学习模型进行训练,最终得到权重参数较好的地质雷达岩溶预报图像的双曲异常检测模型。试验结果及应用案例表明,Mask R-CNN目标检测方法在地质雷达岩溶预报图像的目标检测与识别上取得了良好的效果,有效提高了地质雷达图像的智能化识别效率。

基于改进Mask R-CNN的矿石类型检测算法

针对不同的矿石类型进行选矿工艺参数设置和操作有利于保障精矿质量,提高回收率和降低物资消耗。由于缺乏矿石类型实时检测有效手段,选矿过程的精准操作目前较难实现。为此,本文提出一种全面改进的Mask R-CNN矿石类型检测算法。算法步骤:①使用ResNetV1d-50提取矿石图像各阶段的特征图,并在主干网络中加入可变形卷积以便增强异形矿石的特征;②改进FPN,通过在主干网络的C5特征层加入特征残差模块,并融合到P5特征层,得到具有更强语义信息的多尺度特征图;③改进RPN,设计自适应的正样本IOU选取方案来匹配宽高比异常的矿石,进一步提高异形矿石的识别精度;④在RoIAlign网络基础上加入Global-Context,以提高小矿石的检测能力;⑤在数据增强和训练技巧方面对模型进行改进。结果表明,本文算法的平均精度为67.92%,平均交并比为63.54%,分别比基准模型提高了13.67%和9.71%。本文研究方法在矿石类型识别领域具有较好的应用价值。

X-ray在鱼体组织及微量元素检测中的应用

鱼类硬质组织物,特别是骨骼支持鱼体和保护其体内器官的组织,对其进行特征检测分析是研究鱼类游泳运动、鱼类解剖、鱼体建模等的数据基础。随着X-ray技术的发展和国产设备的广泛应用,其仪器设备成本明显降低,使得X-ray在渔业研究与自动化生产中的应用成为可能。首先介绍了X-ray技术的基本原理与其在鱼体组织检测中的应用,X-ray技术在鱼体组织及微量元素检测中的应用主要分为鱼类组织器官的无损检测和鱼体微量元素检测两部分,其中分别介绍了包括照相法、数字成像法、衍射技术和吸收光谱法等X-ray技术;然后综述其在鱼体组织器官建模、鱼骨检测、鱼类化石研究、鱼耳石分析和鱼体微量元素检测方面的应用,总结了Xray在渔业领域应用中存在的问题;最后对X-ray的渔业应用方向进行展望。

基于Mask R-CNN和迁移学习的无人机遥感影像杉木单木树冠提取

【目的】利用无人机遥感影像对树冠进行自动化提取,获取高精度树冠信息。【方法】该研究提出一种基于Mask RCNN和迁移学习的无人机影像单木树冠提取方法。首先,选用在Faster R-CNN基础上改进优化的Mask R-CNN实例分割模型,特征提取网络在ResNet50残差网络和ResNet101残差网络二者间选取最优。其次,引入迁移学习与Mask RCNN一起训练,联合迁移学习的导向作用降低训练时间,提高训练精度。【结果】Mask R-CNN模型的总体精度为93.59%,用户精度为65.46%,F1分数为76.05%,平均精度均值为0.31;载入迁移学习后的Mask R-CNN模型在同等训练条件下比原模型的用户精度提升29.53%,F1分数提升19.63%,平均精度均值提升0.21;分别以ResNet50和ResNet101为特征提取网络的Mask R-CNN模型中,ResNet50+Mask R-CNN模型的总体精度、用户精度、F1分数、平均精度均值各为96.94%、95.57%、96.17%、0.54,ResNet101+Mask R-CNN模型的总体精度、用户精度、F1分数、平均精度均值各为96.20%、94.41%、95.19%、0.49;其中载入迁移学习的ResNet50+Mask R-CNN模型在预测东西冠幅、南北冠幅、树冠面积与样方郁闭度的预测决定系数分别为0.87、0.84、0.93和0.83。【结论】本研究提出的基于Mask R-CNN和迁移学习的方法得到了较为精准的树冠参数结果,为无人机遥感影像评估树木资源提供了一种快速高效的解决方案。

基于多目标采样和改进Mask R-CNN的木瓜成熟度检测

目的:提高木瓜成熟度检测准确率及鲁棒性。方法:提出一种基于多目标采样和改进Mask R-CNN的木瓜成熟度检测方法。通过均值平均精度、准确率、精确率—召回率曲线和计算时间等指标,验证所提方法的有效性和鲁棒性,并将其检测效果与Faster R-CNN、RetinaNet和CenterMask等方法进行对比。结果:试验方法对木瓜成熟度检测的平均精度均值、50%平均精度均值、75%平均精度均值分别为98.43%,98.67%,98.68%,对未成熟、半成熟和成熟木瓜成熟度的平均检测精度为99.38%,98.81%,99.37%。结论:该方法可用于开发木瓜成熟度检测的电子系统,提升木瓜成熟度检测和木瓜分级的性能。

一种Mask交互融合预训练知识的低资源神经机器翻译方法

融合预训练语言知识是提升低资源神经机器翻译性能的有效手段.现有融合预训练语言知识的方法都比较复杂,计算资源消耗较大.针对以上问题,本文提出了一种简单有效的Mask交互融合预训练知识的低资源神经机器翻译方法,首先利用BERT的句子表征与源语言表征的自注意力交互计算两种表征的相似度,根据相似度值构造Mask知识矩阵,然后将Mask知识矩阵作用于源语言表征,自适应地将BERT表征中对低资源神经机器翻译任务有益的语言知识融入翻译模型,提升翻译模型对语言知识的表征能力.在IWSLT标准低资源翻译任务上的实验结果表明,与Transformer基线模型相比,所提方法获得了0.9~3.39的BLEU值提升,证明了所提方法能够有效利用预训练语言知识增强神经机器翻译性能.

基于改进Mask R-CNN的高密度砂岩颗粒的分割识别

针对高密度颗粒密度大,数量多,形态不一,且颜色相近的情况,通过传统方法对砂岩颗粒分割难度存在检测不准和漏检的不足。想要在少量样本中获取更好的效果,变得更加困难。基于上述问题提出一种基于改进Mask R-CNN的DGC-Mask R-CNN检测模型,针对少量样本、高密度砂岩颗粒的分割与识别。首先收集了128张超高分辨率的图片,每张图片有近200个砂岩颗粒实例,共26200个实例对象。为了使模型拥有更好的泛化能力,防止少量样本下的过拟合,使用Albu进行图像增强。用自监督预训练模型Barlow Twins来对砂岩颗粒的特征进行初步提取。在DGC-Mask R-CNN中,构建ResNet50模型作为骨干特征提取网络,在ResNet50的BottleNeck的C 3、C 4、C 5特征卷积层中改进传统卷积方式,使用可变形卷积神经网络(deformable convolutional neural network,DCN),并添加全局上下文建模框架(global context block,GCB)注意力机制。在上采样器的多个级联上采样模块中,结合改进的上采样算法CARAFE。实验结果表明,改进后的DGC-Mask R-CNN,使得检测与分割识别的平均精度达到88.9%和88.8%,与传统的Mask R-CNN、Cascade-Mask R-CNN、Mask Scoring R-CNN、HybridTaskCascade相比检测精度更高。在均值平均精度方面,与其他模型相比提升较为明显。将模型分割后得到的结果,进行砂岩颗粒的统计以及长短轴的计算,可实现对该部分砂岩颗粒的溯源,计算地壳运动导致的砂岩迁移的距离,进而评估地下油藏。

The Soft X-ray Imager(SXI)on the SMILE Mission

The Soft X-ray Imager(SXI)is part of the scientific payload of the Solar wind Magnetosphere Ionosphere Link Explorer(SMILE)mission.SMILE is a joint science mission between the European Space Agency(ESA)and the Chinese Academy of Sciences(CAS)and is due for launch in 2025.SXI is a compact X-ray telescope with a wide field-of-view(FOV)capable of encompassing large portions of Earth’s magnetosphere from the vantage point of the SMILE orbit.SXI is sensitive to the soft X-rays produced by the Solar Wind Charge eXchange(SWCX)process produced when heavy ions of solar wind origin interact with neutral particles in Earth’s exosphere.SWCX provides a mechanism for boundary detection within the magnetosphere,such as the position of Earth’s magnetopause,because the solar wind heavy ions have a very low density in regions of closed magnetic field lines.The sensitivity of the SXI is such that it can potentially track movements of the magnetopause on timescales of a few minutes and the orbit of SMILE will enable such movements to be tracked for segments lasting many hours.SXI is led by the University of Leicester in the United Kingdom(UK)with collaborating organisations on hardware,software and science support within the UK,Europe,China and the United States.

Global hybrid simulations of soft X-ray emissions in the Earth’s magnetosheath

Earth’s magnetopause is a thin boundary separating the shocked solar wind plasma from the magnetospheric plasmas,and it is also the boundary of the solar wind energy transport to the magnetosphere.Soft X-ray imaging allows investigation of the large-scale magnetopause by providing a two-dimensional(2-D)global view from a satellite.By performing 3-D global hybrid-particle-in-cell(hybrid-PIC)simulations,we obtain soft X-ray images of Earth’s magnetopause under different solar wind conditions,such as different plasma densities and directions of the southward interplanetary magnetic field.In all cases,magnetic reconnection occurs at low latitude magnetopause.The soft X-ray images observed by a hypothetical satellite are shown,with all of the following identified:the boundary of the magnetopause,the cusps,and the magnetosheath.Local X-ray emissivity in the magnetosheath is characterized by large amplitude fluctuations(up to 160%);however,the maximum line-of-sight-integrated X-ray intensity matches the tangent directions of the magnetopause well,indicating that these fluctuations have limited impact on identifying the magnetopause boundary in the X-ray images.Moreover,the magnetopause boundary can be identified using multiple viewing geometries.We also find that solar wind conditions have little effect on the magnetopause identification.The Solar wind Magnetosphere Ionosphere Link Explorer(SMILE)mission will provide X-ray images of the magnetopause for the first time,and our global hybrid-PIC simulation results can help better understand the 2-D X-ray images of the magnetopause from a 3-D perspective,with particle kinetic effects considered.

Finding the magnetopause location using soft X-ray observations and a statistical inverse method

Variability in the location and shape of the dayside magnetopause is attributed to magnetic reconnection,a fundamental process that enables the transfer of mass,energy,and momentum from the solar wind into the magnetosphere.The spatial and temporal properties of the magnetopause,under varying solar and magnetospheric conditions,remain largely unknown because empirical studies using in-situ observations are challenging to interpret.Global wide field-of-view(FOV)imaging is the only means to simultaneously observe the spatial distribution of the plasma properties over the vast dayside magnetospheric region and,subsequently,quantify the energy transport from the interplanetary medium into the terrestrial magnetosphere.Two upcoming missions,ESA/CAS SMILE and NASA’s LEXI will provide wide-field imagery of the dayside magnetosheath in soft X-rays,an emission generated by charge exchange interactions between high charge-state heavy ions of solar wind origin and exospheric neutral atoms.High-cadence two-dimensional observations of the magnetosheath will allow the estimation of dynamic properties of its inner boundary,the magnetopause,and enable studies of its response to changes in the solar wind dynamic pressure and interplanetary magnetic field orientation.This work introduces a statistically-based estimation approach based on inverse theory to estimate the spatial distribution of magnetosheath soft X-ray emissivities and,with this,identify the location of the magnetopause over the Sun−Earth line.To do so,we simulate the magnetosheath structure using the MHD-based OpenGGCM model and generate synthetic soft X-ray images using LEXI’s orbit and attitude information.Our results show that 3-D estimations using the described statistically-based technique are robust against Poisson-distributed shot noise inherent to soft X-ray images.Also,our proposed methodology shows that the accuracy of both three-dimensional(3-D)estimation and the magnetopause standoff distance calculation highly depends on the observational point.

Two methods for separating the magnetospheric solar wind charge exchange soft X-ray emission from the diffuse X-ray background

Solar wind charge exchange(SWCX)is the process of solar wind high-valence ions exchanging charges with neutral components and generating soft X-rays.Recently,detecting the SWCX emission from the magnetosphere is proposed as a new technique to study the magnetosphere using panoramic soft X-ray imaging.To better prepare for the data analysis of upcoming magnetospheric soft X-ray imaging missions,this paper compares the magnetospheric SWCX emission obtained by two methods in an XMM-Newton observation,during which the solar wind changed dramatically.The two methods differ in the data used to fit the diffuse X-ray background(DXB)parameters in spectral analysis.The method adding data from the ROSAT All-Sky Survey(RASS)is called the RASS method.The method using the quiet observation data is called the Quiet method,where quiet observations usually refer to observations made by the same satellite with the same target but under weaker solar wind conditions.Results show that the spectral compositions of magnetospheric SWCX emission obtained by the two methods are very similar,and the changes in intensity over time are highly consistent,although the intensity obtained by the RASS method is about 2.68±0.56 keV cm^(-2)s^(-1)sr^(-1)higher than that obtained by the Quiet method.Since the DXB intensity obtained by the RASS method is about 2.84±0.74 keV cm^(-2)s^(-1)sr^(-1)lower than that obtained by the Quiet method,and the linear correlation coefficient between the difference of SWCX and DXB obtained by the two methods in diffe rent energy band is close to-1,the diffe rences in magnetospheric SWCX can be fully attributed to the diffe rences in the fitted DXB.The difference between the two methods is most significant when the energy is less than 0.7 keV,which is also the main energy band of SWCX emission.In addition,the difference between the two methods is not related to the SWCX intensity and,to some extent,to solar wind conditions,because SWCX intensity typically va ries with the solar wind.In summary,both methods are robust and reliable,and should be considered based on the best available options.

SMILE soft X-ray Imager flight model CCD370 pre-flight device characterisation

Throughout the SMILE mission the satellite will be bombarded by radiation which gradually damages the focal plane devices and degrades their performance.In order to understand the changes of the CCD370s within the soft X-ray Imager,an initial characterisation of the devices has been carried out to give a baseline performance level.Three CCDs have been characterised,the two flight devices and the flight spa re.This has been carried out at the Open University in a bespo ke cleanroom measure ment facility.The results show that there is a cluster of bright pixels in the flight spa re which increases in size with tempe rature.However at the nominal ope rating tempe rature(-120℃) it is within the procure ment specifications.Overall,the devices meet the specifications when ope rating at -120℃ in 6 × 6 binned frame transfer science mode.The se rial charge transfer inefficiency degrades with temperature in full frame mode.However any charge losses are recovered when binning/frame transfer is implemented.

Estimating the subsolar magnetopause position from soft X-ray images using a low-pass image filter

The Lunar Environment heliospheric X-ray Imager(LEXI)and Solar wind Magnetosphere Ionosphere Link Explorer(SMILE)missions will image the Earth’s dayside magneto pause and cusps in soft X-rays after their respective launches in the near future,to specify glo bal magnetic reconnection modes for varying solar wind conditions.To suppo rt the success of these scientific missions,it is critical to develop techniques that extract the magnetopause locations from the observed soft X-ray images.In this research,we introduce a new geometric equation that calculates the subsolar magnetopause position(RS)from a satellite position,the look direction of the instrument,and the angle at which the X-ray emission is maximized.Two assumptions are used in this method:(1)The look direction where soft X-ray emissions are maximized lies tangent to the magnetopause,and(2)the magnetopause surface near the subsolar point is almost spherical and thus RSis nea rly equal to the radius of the magneto pause curvature.We create synthetic soft X-ray images by using the Open Geospace General Circulation Model(OpenGGCM)global magnetohydrodynamic model,the galactic background,the instrument point spread function,and Poisson noise.We then apply the fast Fourier transform and Gaussian low-pass filte rs to the synthetic images to re move noise and obtain accurate look angles for the soft X-ray pea ks.From the filte red images,we calculate RS and its accuracy for different LEXI locations,look directions,and solar wind densities by using the OpenGGCM subsolar magnetopause location as ground truth.Our method estimates RS with an accuracy of<0.3 RE when the solar wind density exceeds>10 cm-3.The accuracy improves for greater solar wind densities and during southward interplanetary magnetic fields.The method ca ptures the magnetopause motion during southwa rd interplaneta ry magnetic field turnings.Consequently,the technique will enable quantitative analysis of the magnetopause motion and help reveal the dayside reconnection modes for dynamic solar wind conditions.This technique will suppo rt the LEXI and SMILE missions in achieving their scientific o bjectives.

Solar wind ion charge state distributions and compound cross sections for solar wind charge exchange X-ray emission

Solar Wind Charge eXchange X-ray(SWCX) emission in the heliosphere and Ea rth’s exosphere is a hard to avoid signal in soft Xray obse rvations of astrophysical targets.On the other hand,the X-ray imaging possibilities offered by the SWCX process has led to an increasing number of future dedicated space missions for investigating the solar wind-terrestrial inte ractions and magnetospheric interfaces.In both cases,accurate modelling of the SWCX emission is key to correctly interpret its signal,and remove it from obse rvations,when needed.In this paper,we compile solar wind abundance measurements from ACE for different solar wind types,and atomic data from literature,including charge exchange cross-sections and emission probabilities,used fo r calculating the compound cross-section a for the SWCX X-ray emission.We calculate a values for charge-exchange with H and He,relevant to soft X-ray energy bands(0.1-2.0 keV)for various solar wind types and solar cycle conditions.

改进Mask R-CNN的无人机影像建筑物提取

从无人机影像中自动提取建筑物对城乡规划和管理至关重要,然而,在复杂背景干扰和建筑物外观变化很大的情况下给实例提取带来挑战。因此,提出一种改进的Mask区域卷积神经网络(R-CNN)方法用于无人机影像的建筑物自动实例提取。改进方法以ResNet-101作为特征提取网络,在特征融合网络方面,通过添加自底向上的路径增强整个特征层次的定位能力,同时在特征融合中加入空洞空间金字塔池化模块(ASPP)来提高多尺度能力与改善模型性能。在自制建筑物数据集上的综合实验结果表明,与原始的Mask R-CNN方法相比,改进方法的mAP值提高了2.6%,能够很好地实现无人机影像建筑物实例提取。

Hybrid-Vlasov simulation of soft X-ray emissions at the Earth’s dayside magnetospheric boundaries

Solar wind charge exchange produces emissions in the soft X-ray energy range which can enable the study of near-Earth space regions such as the magnetopause,the magnetosheath and the polar cusps by remote sensing techniques.The Solar wind Magnetosphere Ionosphere Link Explorer(SMILE)and Lunar Environment heliospheric X-ray Imager(LEXI)missions aim to obtain soft Xray images of near-Earth space thanks to their Soft X-ray Imager(SXI)instruments.While earlier modeling works have already simulated soft X-ray images as might be obtained by SMILE SXI during its mission,the numerical models used so far are all based on the magnetohydrodynamics description of the space plasma.To investigate the possible signatures of ion-kinetic-scale processes in soft Xray images,we use for the first time a global hybrid-Vlasov simulation of the geospace from the Vlasiator model.The simulation is driven by fast and tenuous solar wind conditions and purely southward interplanetary magnetic field.We first produce global X-ray images of the dayside near-Earth space by placing a virtual imaging satellite at two different locations,providing meridional and equatorial views.We then analyze regional features present in the images and show that they correspond to signatures in soft X-ray emissions of mirrormode wave structures in the magnetosheath and flux transfer events(FTEs)at the magnetopause.Our results suggest that,although the time scales associated with the motion of those transient phenomena will likely be significantly smaller than the integration time of the SMILE and LEXI imagers,mirror-mode structures and FTEs can cumulatively produce detectable signatures in the soft X-ray images.For instance,a local increase by 30%in the proton density at the dayside magnetopause resulting from the transit of multiple FTEs leads to a 12%enhancement in the line-of-sight-and time-integrated soft X-ray emissivity originating from this region.Likewise,a proton density increase by 14%in the magnetosheath associated with mirror-mode structures can result in an enhancement in the soft X-ray signal by 4%.These are likely conservative estimates,given that the solar wind conditions used in the Vlasiator run can be expected to generate weaker soft X-ray emissions than the more common denser solar wind.These results will contribute to the preparatory work for the SMILE and LEXI missions by providing the community with quantitative estimates of the effects of small-scale,transient phenomena occurring on the dayside.

On the apparent line-of-sight alignment of the peak X-ray intensity of the magnetosheath and the tangent to the magnetopause,as viewed by SMILE-SXI

The Soft X-ray Imager(SXI)on board the Solar wind Magnetosphere Ionosphere Link Explorer(SMILE)spacecraft will be able to view the Earth’s magnetosheath in soft X-rays.Simulated images of the X-ray emission visible from the position of SMILE are created for a range of solar wind densities by using 3 years of the SMILE mission orbit,together with models of the expected X-ray emissivity from the Earth’s magnetosheath.Results from global magnetohydrodynamic simulations and a simple model for exospheric neutral densities are used to compare the locations of the lines of sight along which integrated soft X-ray intensities peak with the lines of sight lying tangent to surfaces(defined here to be the magnetopause)along which local soft X-ray intensities peak or exhibit their strongest gradients,or both,for strongly southward interplanetary magnetic field conditions when no depletion or low-latitude boundary layers are expected.Where,in the parameter space of the various times and seasons,orbital phases,solar wind conditions,and magnetopause models,the alignment of the X-ray emission peak with the magnetopause tangent is good,or is not,is presented.The main results are as follows.The spacecraft needs to be positioned well outside the magnetopause;low-altitude times near perigee are not good.In addition,there are seasonal aspects:dayside-apogee orbits are generally very good because the spacecraft travels out sunward at high altitude,but nightside-apogee orbits,behind the Earth,are bad because the spacecraft only rarely leaves the magnetopause.Dusk-apogee and dawnapogee orbits are intermediate.Dayside-apogee orbits worsen slightly over the first three mission years,whereas nightside-apogee orbits improve slightly.Additionally,many more times of good agreement with the peak-to-tangent hypothesis occur when the solar wind is in a high-density state,as opposed to a low-density state.In a high-density state,the magnetopause is compressed,and the spacecraft is more often a good distance outside the magnetopause.

基于X-rayμCT技术的玉米干燥损伤定量研究

应力裂纹是玉米干燥损伤的主要形式,一些微裂纹无法肉眼观察但影响玉米的机械强度、发芽率以及活力等。为此,利用X-rayμCT技术对不同干燥条件下玉米进行扫描,使用深度学习算法分割与胚乳密度相近的胚区域并结合软件对玉米内部物性特征参数进行提取,提出了一种新的玉米内部损伤定量评价方法。结合定量研究结果提出的玉米干燥损伤模型,可为玉米损伤预测提供新思路。