Study on Image Recognition Algorithm for Residual Snow and Ice on Photovoltaic Modules

The accumulation of snow and ice on PV modules can have a detrimental impact on power generation,leading to reduced efficiency for prolonged periods.Thus,it becomes imperative to develop an intelligent system capable of accurately assessing the extent of snow and ice coverage on PV modules.To address this issue,the article proposes an innovative ice and snow recognition algorithm that effectively segments the ice and snow areas within the collected images.Furthermore,the algorithm incorporates an analysis of the morphological characteristics of ice and snow coverage on PV modules,allowing for the establishment of a residual ice and snow recognition process.This process utilizes both the external ellipse method and the pixel statistical method to refine the identification process.The effectiveness of the proposed algorithm is validated through extensive testing with isolated and continuous snow area pictures.The results demonstrate the algorithm’s accuracy and reliability in identifying and quantifying residual snow and ice on PV modules.In conclusion,this research presents a valuable method for accurately detecting and quantifying snow and ice coverage on PV modules.This breakthrough is of utmost significance for PV power plants,as it enables predictions of power generation efficiency and facilitates efficient PV maintenance during the challenging winter conditions characterized by snow and ice.By proactively managing snow and ice coverage,PV power plants can optimize energy production and minimize downtime,ensuring a sustainable and reliable renewable energy supply.

A Single Image Derain Method Based on Residue Channel Decomposition in Edge Computing

The numerous photos captured by low-price Internet of Things(IoT)sensors are frequently affected by meteorological factors,especially rainfall.It causes varying sizes of white streaks on the image,destroying the image texture and ruining the performance of the outdoor computer vision system.Existing methods utilise training with pairs of images,which is difficult to cover all scenes and leads to domain gaps.In addition,the network structures adopt deep learning to map rain images to rain-free images,failing to use prior knowledge effectively.To solve these problems,we introduce a single image derain model in edge computing that combines prior knowledge of rain patterns with the learning capability of the neural network.Specifically,the algorithm first uses Residue Channel Prior to filter out the rainfall textural features then it uses the Feature Fusion Module to fuse the original image with the background feature information.This results in a pre-processed image which is fed into Half Instance Net(HINet)to recover a high-quality rain-free image with a clear and accurate structure,and the model does not rely on any rainfall assumptions.Experimental results on synthetic and real-world datasets show that the average peak signal-to-noise ratio of the model decreases by 0.37 dB on the synthetic dataset and increases by 0.43 dB on the real-world dataset,demonstrating that a combined model reduces the gap between synthetic data and natural rain scenes,improves the generalization ability of the derain network,and alleviates the overfitting problem.

Residues of Logarithmic Differential Forms in Complex Analysis and Geometry

In the article, we discuss basic concepts of the residue theory of logarithmic and multi-logarithmic differential forms, and describe some aspects of the theory, de-veloped by the author in the past few years. In particular, we introduce the notion of logarithmic differential forms with the use of the classical de Rham lemma and give an explicit description of regular meromorphic differential forms in terms of residues of logarithmic or multi-logarithmic differential forms with respect to hypersurfaces, com-plete intersections or pure-dimensional Cohen-Macaulay spaces. Among other things, several useful applications are considered, which are related with the theory of holo-nomic D-modules, the theory of Hodge structures, the theory of residual currents and others.

Joint Rain Streaks & Haze Removal Network for Object Detection

In the realm of low-level vision tasks,such as image deraining and dehazing,restoring images distorted by adverse weather conditions remains a significant challenge.The emergence of abundant computational resources has driven the dominance of deep Convolutional Neural Networks(CNNs),supplanting traditional methods reliant on prior knowledge.However,the evolution of CNN architectures has tended towards increasing complexity,utilizing intricate structures to enhance performance,often at the expense of computational efficiency.In response,we propose the Selective Kernel Dense Residual M-shaped Network(SKDRMNet),a flexible solution adept at balancing computational efficiency with network accuracy.A key innovation is the incorporation of an M-shaped hierarchical structure,derived from the U-Net framework as M-Network(M-Net),within which the Selective Kernel Dense Residual Module(SDRM)is introduced to reinforce multi-scale semantic feature maps.Our methodology employs two sampling techniques-bilinear and pixel unshuffled and utilizes a multi-scale feature fusion approach to distil more robust spatial feature map information.During the reconstruction phase,feature maps of varying resolutions are seamlessly integrated,and the extracted features are effectively merged using the Selective Kernel Fusion Module(SKFM).Empirical results demonstrate the comprehensive superiority of SKDRMNet across both synthetic and real rain and haze datasets.

融合Inception V1-CBAM-CNN的轴承剩余寿命预测模型

针对现有的滚动轴承剩余寿命(RUL)预测方法精度低、轴承健康指标(HI)构建困难等问题,提出了一种基于卷积神经网络(CNN)并融合Inception V1模块和卷积注意力机制模块(CBAM)的滚动轴承RUL预测模型。首先,在CNN中添加了CBAM机制,并进行了加权处理,在通道和空间维度对重要特征进行了强化,对次要特征进行了抑制,通过添加改进的InceptionV1模块,提高了CNN通道间信息交互水平,全面提取了退化特征;然后,进行了网络优化,采用全局最大池化(GMP)方法对模型进行了简化,采用Dropout和批量归一化(BN)方法,避免了过拟合,提高了精度,且克服了训练时出现的梯度消失问题;最后,对数据进行了处理,将降噪后的信号重组为三维张量,将其作为HI,构建了退化标签,引入了评价指标,采用PHM2012轴承数据集进行了实验验证,在3种工况下将其与深度神经网络(DNN)、CNN方法、结合注意力机制的残差网络方法(ResNet)进行了对比。研究结果表明:该方法在变负载条件下的平均RMSE为0.033,较其他方法的RMSE值分别降低了86%、78%和69%,在预测精度和泛化能力方面具有明显优势。

FIBTNet:Building Change Detection for Remote Sensing Images Using Feature Interactive Bi-Temporal Network

In this paper,a feature interactive bi-temporal change detection network(FIBTNet)is designed to solve the problem of pseudo change in remote sensing image building change detection.The network improves the accuracy of change detection through bi-temporal feature interaction.FIBTNet designs a bi-temporal feature exchange architecture(EXA)and a bi-temporal difference extraction architecture(DFA).EXA improves the feature exchange ability of the model encoding process through multiple space,channel or hybrid feature exchange methods,while DFA uses the change residual(CR)module to improve the ability of the model decoding process to extract different features at multiple scales.Additionally,at the junction of encoder and decoder,channel exchange is combined with the CR module to achieve an adaptive channel exchange,which further improves the decision-making performance of model feature fusion.Experimental results on the LEVIR-CD and S2Looking datasets demonstrate that iCDNet achieves superior F1 scores,Intersection over Union(IoU),and Recall compared to mainstream building change detectionmodels,confirming its effectiveness and superiority in the field of remote sensing image change detection.

考虑注射成型特性影响的汽车前端模块结构性能仿真

利用Autodesk Helius PFA软件,将注射成型纤维取向张量和残余应力及应变分布等历史映射至前端模块的结构性能评价仿真。结果表明,注射成型纤维取向张量和残余应力及应变分布等成形历史影响前端模块的服役性能评价。与试验测试结果相比,映射和未映射注射成形历史的仿真最大位移均偏小。相比之下,映射注射成型纤维取向张量和残余应力及应变分布等成形历史的最大位移较大,且这种趋势随变形位移的增大而增大。当试验测试位移很小时,两种仿真方法的结果都与试验有较大的误差。当试验测试位移大于0.65 mm时,映射注射成型历史的百分误差Er、极差R和方差s2均较小,有利于指导玻璃纤维增强塑料车身结构的精准设计。

改进YOLOv4的遥感图像目标检测算法

为有效解决遥感图像目标检测算法在复杂背景下的检测效果不佳的问题,提出一种改进YOLOv4的目标检测算法。设计一种跨阶段残差结构,替换原主干网络的简单残差结构,降低模型参数量和计算负担;引入CBAM注意力机制,加强CSP模块间有效特征交互;使用跨阶段分层卷积模块重构特征融合阶段对深层特征图的处理方式,防止网络退化和梯度消失;采用Mish激活函数,增强融合网络对非线性特征的提取能力。在RSOD、DIOR数据集上的实验结果表明,改进YOLOv4算法的测试mAP相比原YOLOv4算法分别高出4.5%、7.3%,其检测速度分别达到48 fps、45 fps,在保证实时性的同时检测精度有较大提升。

基于scSE非局部双流ResNet网络的行为识别

针对双流网络对包含冗余信息的视频帧存在识别率低的问题,在双流网络的基础上引入scSE(Spatial and Channel Squeeze&Excitation Block)和非局部操作,构建SC_NLResNet行为识别框架。该框架将视频划分为等分不重叠的时序段并在每段上稀疏采样,提取RGB帧以及光流图作为scSE模块的输入;将经过scSE处理的特征输入非局部双流ResNet网络中,融合各分段得到最终的预测结果。在UCF101以及Hmdb51数据集上实验准确率分别达到96.9%和76.2%,结果表明,非局部操作与scSE模块结合可以增强特征时空上以及通道间的信息提高准确率,验证了SC_NLResNet网络的有效性。

基于纹理先验的扩张残差注意力相似性去噪网络

目前,大多数基于卷积神经网络的图像去噪模型不能充分利用图像数据的冗余性,这限制了模型的表达能力。而且,为了有效去噪,往往将边缘信息用作先验知识,而纹理信息通常被忽略。针对这些问题,提出一种新的图像去噪网络,该网络首先使用注意力相似性模块提取图像的全局相似性特征,通过平均池化来平滑和抑制注意力相似性模块中的噪声,以进一步提高网络性能;其次使用扩张残差模块来提取图像的局部和全局特征;最后使用全局残差学习增强网络从浅层到深层的去噪效果。此外,还设计一种纹理提取网络从噪声图像中提取局部二值模式以获取纹理信息,利用纹理信息作为先验知识,可在去噪过程中保留演化图像中的细节。实验结果表明,与一些先进的去噪网络相比,新提出的去噪网络在图像视觉上有很大改善、效率更高且峰值信噪比提高了2 dB左右,结构相似性提高了3%左右,更有利于实际应用。

“三调针法”治疗脊髓损伤后神经源性膀胱临床观察

目的:观察“三调针法”治疗脊髓损伤后神经源性膀胱(neurogenic bladder,NB)的临床疗效。方法:选择脊髓损伤后NB患者60例,随机分为对照组和观察组,每组30例。对照组采用间歇无菌导尿、训练辅助排尿等常规治疗,观察组在对照组基础上采用“三调针法”进行治疗,比较两组治疗前后尿流动力学、残余尿量的变化情况及临床疗效。结果:治疗后两组尿流动力学指标均明显改善(P<0.05),观察组较对照组改善明显(P<0.05)。治疗后两组膀胱残余尿量均较治疗减少(P<0.05);观察组较对照减少明显(P<0.05);总有效率观察组[100.00%(30/30)]高于对照组[83.33%(25/30)](P<0.05)。结论:“三调针法”治疗脊髓损伤后NB,可改善尿流动力学,恢复排尿功能,提高临床疗效。

基于残差卷积网络的多传感器融合永磁同步电机故障诊断

作为工业生产与日常生活的常见设备,永磁同步电机的故障诊断研究具有十分重要的意义。以永磁同步电机的匝间短路、退磁、轴承故障为诊断目标,提出一种新型的多传感器特征融合网络(MSFFN),结合多传感器融合技术与卷积神经网络实现永磁同步电机的可靠故障诊断。网络采用2个带有残差模块的卷积神经网络,对输入的电流信号与振动信号并行提取隐藏特征,并设计一种中间特征融合模块(IFFM)有效融合电流和振动的各层隐藏特征,IFFM基于注意力机制对网络中的电流特征与振动特征进行筛选,自适应关注不同信号的内在相关特征,以实现更好的诊断效果。搭建了故障样机测试平台进行数据采集与实验验证,实验结果表明,提出方法具有更高的诊断准确率,同时在叠加了强噪声的条件下,具备更强的抗干扰能力。

基于双阶段特征解耦网络的单幅图像去雨方法

针对现有的单幅图像去雨方法无法有效地平衡背景图像细节恢复与有雨分量去除问题,提出一种基于双阶段特征解耦网络的单幅图像去雨方法,采用渐进式的学习方式从粗到细分阶段进行单幅图像去雨.首先构建压缩激励残差模块,实现背景图像与有雨分量的初步分离;然后设计全局特征融合模块,其中特别引入特征解耦模块分离有雨分量和背景图像的特征,实现细粒度的图像去雨;最后利用重构损失、结构相似损失、边缘感知损失和纹理一致性损失构成的复合损失函数训练网络,实现高质量的无雨图像重构.实验结果表明,在Test100合成雨图数据集上,所提方法峰值信噪比为25.57dB,结构相似性为0.89;在100幅真实雨图上,所提方法的自然图像质量评估器为3.53,无参考图像空间质量评估器为20.16;在去雨后的RefineNet目标分割任务中,平均交并比为29.41%,平均像素精度为70.06%;视觉效果上,该方法能保留更多的背景图像特征,有效地辅助下游的目标分割任务的开展.

面向超分辨率重建的层次间局部特征增强网络

基于卷积神经网络的超分辨率重建模型以单项传播为主,层次越靠后感知信息的能力越微弱,导致层次间局部特征部分丢失,难以实质提升网络的特征表达能力。针对此问题,提出层次间局部特征增强网络。该方法由级联残差模块、层次间特征增强块和特征感知注意力机制组成。级联残差模块通过有效残差连接增加对残差分支信息的利用;层次间特征增强块提取不同深度特征的依赖关系,自适应调整中间层特征权值增强捕获关键信息的能力;特征感知注意力机制采用方向感知和位置判断的方式准确定位和识别感兴趣对象。多项标准数据集的实验结果表明,该方法能改善超分辨率的视觉重建效果,整体性能优于现有方法。

改进D-Linknet的眼底视网膜血管分割

临床医生可通过观察眼底视网膜血管及其分支对人体是否患有疾病进行早期诊断,但由于视网膜中的血管错综复杂,模型在分割时会出现对微细血管分割精确度不足的问题。为此,提出一种结合残差模块Res2-net以及高效通道注意力机制(efficient channel attention,ECA)的D-Linknet模型。首先,利用Res2-net代替基础模型中的残差模块Res-net以提升每个网络层的感受野;其次,在Res2-net中添加一种结合压缩激励(squeeze and excitation,SE)和门通道(gated channel transformation,GCT)的注意力机制模块,改善处于复杂背景下的血管分割效果和效率;在网络的解码层加入ECA确保模型计算的性能,避免因降维导致的精度下降;最后,融合改进的模型输出图与掩膜图细化分割结果。在公开数据集DRIVE、STARE上进行分割实验,模型准确度(accuracy,AC)分别为97.11%、96.32%,灵敏度(sensitivity,SE)为84.55%、83.92%,曲线下方范围的面积(area under curve,AUC)为0.9873和0.9766,分割效果优于其他模型。实验证明了算法的可行性,为后续研究提供科学依据。

基于改进AOD-Net的图像去雾算法

为了更好解决图像去雾后颜色失真、去雾不彻底和耗时等问题,提出了一种基于改进AOD-Net的图像去雾算法。首先,在原有的卷积模块中引入残差连接,并保留了第二个特征融合层第一层的特征信息,以增强网络的特征提取能力。其次,在第三个特征融合层后引入注意力模块,强化雾图中的关键特征信息,抑制无关背景干扰。最后,采用新的复合损失函数进行训练。实验结果表明,改进算法在公共数据集上的峰值信噪比提高了3.8 dB,结构相似性达到了93.6%。相较于其他去雾算法,该算法在去雾精度和处理效率方面均表现出色。

基于GRU-DRSN的双通道人体活动识别

人体活动识别(human activity recognizition, HAR)在医疗、军工、智能家居等领域有很大的应用空间。传统机器学习方法特征提取难度较大且精度不高。针对上述问题并结合传感器时序特性,提出了一种融合CBAM(convolutional block attention module)注意力机制的GRU-DRSN双通道并行模型,有效避免了传统串行模型因网络深度加深引起梯度爆炸和消失问题。同时并行结构使得两条支路具有相同的优先级,使用深度残差收缩网络(deep residual shrinkage network, DRSN)提取数据的深层空间特征,同时使用门控循环结构(gated recurrent unit, GRU)学习活动样本在时间序列上的特征,同时进行提取样本不同维度的特征,并通过CBAM模块进行特征的权重分配,最后通过Softmax层进行识别,实现了端对端的人体活动识别。使用公开数据集(wireless sensor data mining, WISDM)进行验证,模型平均精度达到了97.6%,与传统机器学习模型和前人所提神经网络模型相比,有更好的识别效果。

一种基于多跳注意残差网络的调制识别算法

为了进一步提升通信信号调制识别的准确率,在ResNet网络的基础上提出一种基于多跳注意残差网络(Multi-skip Attention Residual Network,MARN)的调制识别方法。该方法利用提取不同特征的卷积核进行多跳连接构建3种残差块,进而构建多跳残差网络,提取信号的时域特征;加入CBAM(Convolutional Block Attention Module)注意力机制自适应地调整通道权重,加强信号特征的表征能力;采用自适配归一化(Switchable Normalization,SN)加速网络收敛;加入丢弃率为0.3的AlphaDropout层,提高算法的拟合能力,最终实现对通信信号端到端的分类识别。在RadioML2018.01a数据集上仿真实验,结果表明在信噪比为-10~15 dB下,MARN网络平均识别率达到63.3%,较ResNet网络的平均识别率提升3.7%。

基于卷积—反残差和组合注意力机制的航天器多余物检测

航天器密闭电子设备内腔多余物给航天器飞行安全带来了巨大隐患。由于多余物体积小、与设备内常规组件形态结构相似且易被其他组件遮挡,采用现有的方法对其进行检测时误检、漏检频发。为解决上述问题,提出一种基于卷积—反残差和组合注意力机制的航天器密闭电子设备多余物检测网络RPDN。首先,网络通过构建卷积—反残差模块,保证了多余物细粒度特征的完整性;其次,设计组合注意力机制,增强了多余物特征的表征能力;最后,结合多尺度特征融合模块与目标检测层从多维度进行目标预测。实验结果表明RPDN在各项评价指标上均取得了良好的效果,mAP达到92.16%,检测效率达到了13FPS,实现了航天器密闭电子设备内腔多余物高效、精准检测。

基于轻量级UNet的复杂背景字符语义分割网络

针对传统复杂背景字符分割算法的不足,提出了一种基于轻量级UNet的复杂背景字符语义分割网络.网络结构基于UNet,在特征提取模块中,将传统卷积变为深度可分离卷积,减少了网络特征提取模块的参数量以及计算量,并引入残差学习模块解决网络退化问题.在自制数据集以及H-DIBCO2018公开数据集上展开实验,并与FCN8s、AttationUNet和UNet进行比较.实验结果表明:所提出的网络可同时兼顾计算效率与分割精度,具有实用性.