An Approach to Fault Diagnosis of Rotating Machinery Using the Second-Order Statistical Features of Thermal Images and Simplified Fuzzy ARTMAP

Thermal image, or thermogram, becomes a new type of signal for machine condition monitoring and fault diagnosis due to the capability to display real-time temperature distribution and possibility to indicate the machine’s operating condition through its temperature. In this paper, an investigation of using the second-order statistical features of thermogram in association with minimum redundancy maximum relevance (mRMR) feature selection and simplified fuzzy ARTMAP (SFAM) classification is conducted for rotating machinery fault diagnosis. The thermograms of different machine conditions are firstly preprocessed for improving the image contrast, removing noise, and cropping to obtain the regions of interest (ROIs). Then, an enhanced algorithm based on bi-dimensional empirical mode decomposition is implemented to further increase the quality of ROIs before the second-order statistical features are extracted from their gray-level co-occurrence matrix (GLCM). The highly relevant features to the machine condition are selected from the total feature set by mRMR and are fed into SFAM to accomplish the fault diagnosis. In order to verify this investigation, the thermograms acquired from different conditions of a fault simulator including normal, misalignment, faulty bearing, and mass unbalance are used. This investigation also provides a comparative study of SFAM and other traditional methods such as back-propagation and probabilistic neural networks. The results show that the second-order statistical features used in this framework can provide a plausible accuracy in fault diagnosis of rotating machinery.

Simulation analysis of evolution of hot-images induced by coplanar multi-scatterers

Based on the diffraction theory model for hot-image formation, the evolution of hot-images induced by multiscatterers located in the same plane perpendicular to the propagation axis is numerically simulated. The simulation results show that hot-images induced by coplanar multi-scatterers are also coplanar no matter whether they exist simultaneously or severally. However, if they exist simultaneously the peak intensity of the primary hot-images is weaker than if they exist severally. The unequal competition for energy between the scattered beams from the scatterers leads to the fact that part of the corresponding hot-images are relatively enhanced and the others are restrained. The results show that the hot-images of certain scatterers become stronger when any of these parameters, i.e. amplitude modulation coefficient, phase modulation coefficient and size of the surrounding scatterer, decrease.

A Compound Algorithm of Denoising Using Second-Order and Fourth-Order Partial Differential Equations

In this paper, we propose a compound algorithm for the image restoration. The algorithm is a convex combination of the ROF model and the LLT model with a parameter function 0. The numerical experiments demonstrate that our compound algorithm is efficient and preserves the main advantages of the two models. In particular, the errors of the compound algorithm in L2 norm between the exact images and corresponding restored images are the smallest among the three models. For images with strong noises, the restored images of the compound algorithm are the best in the corresponding restored images. The proposed algorithm combines the fixed point method, an improved AMG method and the Krylov acceleration. It is found that the combination of these methods is efficient and robust in the image restoration.

Channel attention based wavelet cascaded network for image super-resolution

Convolutional neural networks(CNNs) have shown great potential for image super-resolution(SR).However,most existing CNNs only reconstruct images in the spatial domain,resulting in insufficient high-frequency details of reconstructed images.To address this issue,a channel attention based wavelet cascaded network for image super-resolution(CWSR) is proposed.Specifically,a second-order channel attention(SOCA) mechanism is incorporated into the network,and the covariance matrix normalization is utilized to explore interdependencies between channel-wise features.Then,to boost the quality of residual features,the non-local module is adopted to further improve the global information integration ability of the network.Finally,taking the image loss in the spatial and wavelet domains into account,a dual-constrained loss function is proposed to optimize the network.Experimental results illustrate that CWSR outperforms several state-of-the-art methods in terms of both visual quality and quantitative metrics.

基于Hot-Target图和特征边缘保持的图像收缩方法

图像收缩是缩小高分辨率图像以适应不同纵横比小尺寸显示屏幕的过程,关键是收缩后能够凸显图像重要区域,保持连续,避免扭曲.提出一种新的图像收缩方法,该方法首先基于能量失真约束,迭代收缩覆盖图像的四边形网格至目标大小,然后映射,插值目标网格实现图像收缩.能量失真反映了对重要区域的凸显程度、结构的保持效果以及扭曲避免情况,失真越小,目标图像越理想.在该约束下,构成网格的子四边形非均匀收缩,重要度大的收缩小.为准确计算子四边形的重要度,根据图像显著度和边缘构建反映图像重要度的Hot-Target图.最后,通过保持图像直线边,称为特征边缘,避免非均匀收缩引起的边缘扭曲.为提高效率,降低复杂度,该方法由迭代求解线性方程实现.实验结果验证了方法的有效性.

Lensless ghost imaging through the strongly scattering medium

Lensless ghost imaging has attracted much interest in recent years due to its profound physics and potential applications. In this paper we report studies of the robust properties of the lensless ghost imaging system with a pseudo-thermal light source in a strongly scattering medium. The effects of the positions of the strong medium on the ghost imaging are investigated. In the lensless ghost imaging system, a pseudo-thermal light is split into two correlated beams by a beam splitter. One beam goes to a charge-coupled detector camera, labeled as CCD2. The other beam goes to an object and then is collected in another charge-coupled detector camera, labeled as CCD1, which serves as a bucket detector. When the strong medium, a pane of ground glass disk, is placed between the object and CCD1, the bucket detector, the quality of ghost imaging is barely affected and a good image could still be obtained. The quality of the ghost imaging can also be maintained, even when the ground glass is rotating, which is the strongest scattering medium so far. However, when the strongly scattering medium is present in the optical path from the light source to CCD2 or the object, the lensless ghost imaging system hardly retrieves the image of the object. A theoretical analysis in terms of the second-order correlation function is also provided.

Micro-Scanning Error Correction Technique for an Optical Micro-Scanning Thermal Microscope Imaging System

An error correction technique for the micro-scanning instrument of the optical micro-scanning thermal microscope imaging system is proposed. The technique is based on micro-scanning technology combined with the proposed second-order oversampling reconstruction algorithm and local gradient image reconstruction algorithm. In this paper, we describe the local gradient image reconstruction model, the error correction technique, down-sampling model and the error correction principle. In this paper, we use a Lena original image and four low-resolution images obtained from the standard half-pixel displacement to simulate and verify the effectiveness of the proposed technique. In order to verify the effectiveness of the proposed technique, two groups of low-resolution thermal microscope images are collected by the actual thermal microscope imaging system for experimental study. Simulations and experiments show that the proposed technique can reduce the optical micro-scanning errors, improve the imaging effect of the system and improve the system's spatial resolution. It can be applied to other electro-optical imaging systems to improve their resolution.

基于轻量化YOLOv5s的光伏热斑检测定位方法

针对目前目标检测技术在检测光伏热斑效应时模型检测速度低、计算复杂、模型结构复杂等问题,提出基于轻量化YOLOv5s的光伏热斑检测定位方法。首先,以YOLOv5s为基础模型,引入轻量网络ShuffleNetV2改进YOLOv5s的主干网络,利用其分组卷积和通道混洗的设计思想,减少模型参数和计算量,同时保持较高的准确率。其次,引入轻量级卷积GSConv优化YOLOv5s的Neck部分,利用其深度可分离卷积结合标准卷积的形式,降低计算复杂度,优化整体模型。最后利用数据集进行验证。结果表明,轻量化后的模型在保证较高精度的前提下,能够提高检测速度、减少参数量和计算量。

基于红外线热成像技术探索“按之则热气至”的按压次数-热效应机制研究

背景推拿量效关系的客观测量方法越来越多,但将红外线热成像技术与推拿量效关系结合的研究较少。目的通过对比不同的叠掌按压法次数(3、5、7次)下治疗区域温度值、肌张力、压痛阈值变化,为红外线热成像技术运用于推拿量效关系提供依据。方法于2021-06-17—12-24在广东省中医院大德路总院推拿科招募18例胸椎小关节紊乱(病变节段在T3~T4)患者。将患者分为按压7次组(n=6)、按压5次组(n=6)、按压3次组(n=6),接受叠掌按压法治疗,次数分别为7、5、3次。采用红外线热成像仪获取患者全身的红外热像图,提取病变区域的平均温度值;运用软组织张力测试分析系统测量病变胸椎处旁竖脊肌的肌张力;运用痛阈测定仪测量病变胸椎棘突旁体表部位的压痛阈值。结果按压3次组干预后的T3棘突右旁开0.5寸(T3 R)温度值及T4棘突右旁开0.5寸(T4 R)温度值均低于干预前(P<0.05)。干预后,三组T3棘突左旁开0.5寸(T3 L)在0.2 kg的力下肌张力、T3 R在0.2 kg的力下肌张力、T4棘突左旁开0.5寸(T4 L)在0.2 kg的力下肌张力、T4 L在0.5 kg的力下肌张力、T4 L在0.8 kg的力下肌张力、T4 R在0.2 kg的力下肌张力、T4 R在0.5 kg的力下肌张力比较,差异均有统计学意义(P<0.05)。按压7次组干预后T3 R在0.2、0.5 kg的力下肌张力高于干预前(P<0.05)。按压3次组干预后T3 L、T3 R、T4 L、T4 R压痛阈值低于干预前(P<0.05)。按压7次组干预后T3 L、T3 R压痛阈值低于干预前(P<0.05)。三组患者在接受治疗时均未出现不良事件。结论按压次数较少可能产生较好的疗效。红外线热成像技术作为推拿量效关系的测量方式仍需要进一步研究。结合本研究的问题可进一步为红外线热成像与推拿的关系提供研究思路。

基于ECA-ResNet的热轧钢表面缺陷在线识别

基于视觉注意力机制对ResNet模型进行改进,设计了ECA-ResNet网络模型,对热轧钢表面缺陷进行在线识别。该模型在提高检测精度的同时,其推理速度也符合热轧钢生产线对实时性的要求。实验结果表明,ECA-ResNet模型的在线识别精度为98.1%,比经典ResNet网络提高了3.9个百分点。消融实验和对比实验说明,融合ECA注意力模块的ResNet网络的综合性能优于融合CBAM或SE注意力模块的ResNet网络,也优于经典ResNet网络、VGG-16网络和GoogLeNet网络,符合热轧钢表面缺陷的在线识别要求,具有良好的应用前景。

光伏组件热斑效应智能化检测的综述及展望

针对目前光伏组件热斑检测方式较为众多繁杂的问题,总结并系统剖析现存的国内外各种光伏组件传统及智能化的热斑检测方式,将深度学习算法运用于此的最新进展,并着重研究多位研究者对于将各种神经网络、注意力机制以及目标检测模型运用于热斑红外图像的检测上,如多尺度卷积神经网络、基于SpotFPN多尺度特征学习模块等。之后针对光伏组件热斑检测的智能化诊断技术进行实验对比,最后对当前问题展开探讨以及对未来该技术进展的展望。

基于灰度直方图自适应B样条曲线拟合的光伏组件热斑检测方法

红外热图像探测热斑的过程中,会受到复杂环境以及采集过程引入的诸多因素干扰,导致常规图像处理方法无法准确检测出热斑。提出一种改进的热斑检测方法,该方法根据灰度直方图特征进行自适应节点配置的B样条拟合,从而实现阈值选取和热斑分割,不仅同样具有抑制噪声的特点,且提高了热斑检测的准确率。实验证明该方法能实现含有大量噪声的光伏组件红外热图像的热斑检测,提高热斑检测的效率和准确率。

基于卷积神经网络算法的光伏组件热斑图像检测方法研究

热斑效应是造成光伏组件损坏的主要原因之一,为提早发现光伏组件热斑效应,利用卷积神经网络算法优化设计光伏组件热斑图像检测方法;根据光伏组件的工作原理以及热斑的产生机理,设置光伏组件热斑图像检测标准;通过红外热像仪和探测器采集光伏组件热斑图像,并对采集得到的图像展开预处理;利用卷积神经网络算法从轮廓和颜色两个方面提取预处理后热斑图像的特征,根据特征匹配结果得出光伏组件热斑图像的检测结果;实验结果表明:与传统检测方法相比,优化设计方法的漏检率和误检率明显更低,且光伏组件热斑面积检测误差较低,说明其检测效果较好。

基于红外热成像检测的光伏电站异常分析

红外热成像检测具有非接触、检测速度快、检测范围广、易于实现自动化和可以实时观测等优点,在众多行业中都有广泛应用。为了更好地把红外热成像技术应用于光伏电站检测领域,利用手持式红外热成像仪和无人机搭载红外热成像仪两种检测方式,对多个在运光伏电站现场进行了红外热成像检测,根据红外热成像检测结果挑选出光伏电站常见的故障案例,然后分析了这些故障导致的异常红外热成像图的典型特征及形成原因。通过红外热成像检测结果发现:光伏组件和电气设备存在一些热异常问题,导致这些热异常的故障类型可分为5类,分别为:1)光伏组件中单片太阳电池热斑;2)光伏组件内太阳电池串开路;3)光伏组串(或光伏组件)短路或开路;4)光伏阵列完整性缺失或机械受损;5)电气设备异常。研究结果可以为红外热成像技术在光伏电站检测中的应用提供参考和借鉴,通过红外热成像图来准确识别光伏电站的故障类型,及时发现光伏电站存在的发热故障和隐患,从而可以提高光伏电站的运维效率,对促进光伏电站的安全高效运行和新能源行业的发展具有重要意义。

基于主成分分析的光伏热斑红外图像混合噪声去噪方法

为了解决光伏板热斑故障检测时受噪声影响的红外图像分辨率低而导致热斑区域难以识别的问题,提出一种基于主成分分析的红外图像混合噪声自适应去噪方法。该方法通过自适应窗口预处理算法将获取的热斑红外图像进行初步去噪,滤除图像中的低密度椒盐噪声,减小噪声信号对后续选取降噪训练集时所造成的影响;然后,采用基于块匹配的主成分分析法对预处理后的图像信息进行降维处理,提取信号的主要特征,降低噪声滤除时的计算复杂度;最后,使用线性最小均方误差估计对图像进行二次去噪处理,滤除残余噪声;此外,在二次去噪之前重新计算图像噪声水平,使最终的去噪图片获得了更好的视觉效果。实验结果表明:该方法能够有效去除光伏热斑红外图像中的混合噪声,客观评价指标显示噪声较小时,图像结构相似性可保持在0.9,在高密度噪声影响下,峰值信噪比相较于修正的阿尔法均值滤波算法平均提高2 dB,实际视觉效果中保留了图像细节特征,可以明显观测到热斑区域。

基于改进K均值聚类的光伏板缺陷检测方法

为了能够对光伏组件热斑部分准确地识别和提取,提出了一种基于HSV空间模型的改进K均值聚类图像处理方法。首先,将红外图像进行HSV空间转换和双边滤波处理,去除噪声并提高图像对比度;其次,使用高斯核函数估计实现图像灰度概率密度函数提取,并以此获取初始聚类中心;最后,利用先验知识对图像进行K均值聚类,提取和量化热斑缺陷。研究结果表明,该方法能够快速地检测定位热斑位置并统计出光伏板损坏程度,具有较高的精度以及较好的灵敏性和稳定性。

Motion de-blurring by second-order intensity-correlated imaging

For a Hanbury Brown and Twiss system, the influence of relative motion between the object and the detection plane on the resolution of second-order intensity-correlated imaging is investigated. The analytical results, which are backed up by experiments, demonstrate that the amplitude and mode of the object＇s motion have no effect on the second-order intensity-correlated imaging and that high-resolution imaging can be always achieved by using a phase-retrieval method from the diffraction patterns. The use of motion de-blurring imaging for this approach is also discussed.

Direct prejudgement of hot images with detected diffraction rings in high power laser system

A direct prejudgement strategy that takes the diffraction ring as the analysis target is put forward to predict hot images induced by defects of tens of microns in the main amplifier section of high power laser systems. Analysis of hot-image formation process shows that the hot image can be precisely calculated with the extracted intensity oscillation of the diffraction ring on the front surface of the nonlinear plate. The gradient direction matching （GDM） method is adopted to detect diffraction tings. Recognition of simulated diffraction rings shows that it is feasible to directly prejudge hot images induced by those closely spaced defects and the defects that are far apart from each other. Image compression and cluster analysis are utilized to optimize the performance of the GDM method in recognizing actually collected diffraction images. Results show that hot images induced by defects of tens of microns can be directly prejudged without redundant information.

旅游凝视理论下枣庄仙坛山温泉小镇形象建构研究

基于“旅游凝视”理论,运用访谈法与网络文本分析法,分别从东道主与游客双重视角,深入分析枣庄仙坛山温泉小镇的形象建构作用,以期推动案例的形象建设,并对温泉旅游目的地形象建设提供借鉴。研究认为,东道主通过旅游凝视物、媒体表征传播、反凝视旅游者感知,共同建构旅游地形象;旅游者通过凝视景观,以亲身体验积累认知印象与情感印象,形成整体形象,将凝视记忆利用媒介传播出去,综合完成对旅游地形象的建构与再建构;东道主的“反凝视”与游客对旅游目的地的“再建构”形成双向互动,共同作用于旅游目的地形象。

基于改进的ResNet50网络的光伏热斑识别算法

为提高样本在分布不均衡情况下的识别精度,提出一种改进的ResNet50卷积神经网络光伏热斑识别算法。首先,为增加初期红外纹理信息流入、调整网络宽度,设计一种头部分组特征提取模块,并将其嵌入到残差网络中,提高网络在图像细微特征方面的提取能力;然后,将通道注意力机制与残差模块相结合,增加网络通道间的热斑特征信息权重,提高模型识别性能和网络收敛速度;最后,通过图像转换HSV颜色空间、平均H分量梯度直方图峰值等数据预处理方法,将负样本转为多分类数据集,并用于热斑识别网络模型,实现热斑识别结果的可视化。实验结果表明,对比其他算法,改进后的ResNet50网络在识别精度上得到显著提高。