无人机检测光伏组件热斑效应的设计与实现

选用大疆inspire 1无人机作为飞行平台,搭载FLRE Vue机载热像仪机芯,并选用三星的S3C6410处理器作为嵌入式终端,成功实现了自动化检测热斑效应;通过开发软件对出现热斑效应的组件进行分析处理,得出了具体的热斑检测报告。通过无人机检测光伏组件热斑,大幅提高了电站红外热斑检测效率。     

晶体硅光伏组件热斑失效问题研究

以不同反向漏电流等级的多晶硅太阳电池封装成的光伏组件和实际发生热斑失效的光伏组件为研究对象,通过数值模拟和实验研究的方法,对晶体硅光伏组件热斑失效的机理和规律进行理论分析与实验验证。研究结果表明:在阴影遮挡环境下被遮挡组件区域温度和太阳电池反向电流成正向相关性,即反向漏电流越大,组件温度越高;实验同时发现即使在完全无阴影遮挡的情况下,光伏组件也可能因组件封装过程中存在虚焊、空焊等接触不良连接点,形成微小间隙,引发电弧效应,从而导致严重的热斑失效。     

局部阴影下光伏阵列结构优化

为提高光伏阵列发电效率,避免发生热斑效应,通过改进空间向量模型,对局部阴影下光伏组件间的匹配度进行计算,并将匹配度高的光伏组件进行串联,实现对光伏阵列结构的优化。对比分析表明,该算法可提高光伏阵列输出功率,提高组件间的匹配度,避免因不匹配而造成的热斑问题。     

局部遮挡时光伏阵列失配特性建模与分析

结合局部遮挡时太阳电池的等效电路,建立太阳电池发热现象的数学模型,对不同状态下太阳电池发热的主要原因进行分类分析;基于对局部遮挡时光伏组件工作特性的分析,建立以分段函数描述的光伏组件输出特性模型,最后结合实验及仿真结果分析失配引起的输出多峰值特性和热斑效应现象,提出解决太阳电池失配问题的控制方法。为改进光伏系统设计、优化光伏系统控制方法提供理论依据。     

热斑晶硅光伏组件I-V曲线分析及特征模拟研究

基于大量的具有热斑缺陷的晶硅光伏组件测试结果,将热斑种类归纳为阴影遮挡型热斑和非遮挡型热斑。分析阴影遮挡型热斑组件的I-V曲线特征及原因,发现通过I-V曲线阶梯处线段的斜率k可表征热斑电池的漏电流大小,并结合组件EL测试结果分析了非遮挡型热斑产生的原因。解释影响光伏组件漏电流大小的原因,通过Simulink建立光伏组件仿真模型,验证热斑组件I-V曲线特征与热斑电池片漏电水平的关系,提出通过给正常电池片并联电阻来模拟曲线斜率特征,并给出并联电阻的选择依据,并通过设计实验以k值差异为判断标准验证了该模拟方法的准确性。     

晶体硅光伏组件的热斑效应详解

根据晶体硅光伏组件热斑耐久试验的结果,分析太阳电池发热的原因,并设计实验寻找热斑效应影响程度与遮挡面积大小及外接负载大小的关系,最后阐述了目前利用旁路二极管减小热斑效应影响的原理及光伏电站设计运维的注意事项。     

简述光伏组件热斑耐久试验的主要步骤及其原理

光伏组件的热斑现象较为普遍,但IEC等相关标准仅对光伏组件热斑耐久试验的步骤进行了介绍,导致初步接触该试验的人员很难理解采用这些步骤的原因.基于此,在简述热斑耐久试验主要步骤的同时,揭示了各个操作步骤的原理.将光伏组件采用的太阳电池根据其特性分为"A类太阳电池"和"B类太阳电池",其中,对于A类太阳电池而言,最严重的热...     

基于I-V曲线的光伏组件热斑测试与分析

通过对部分热斑组件与正常组件进行红外热成像测试及I-V测试,并对I-V曲线特定区间进行线性拟合分析,确定热斑在I-V曲线上的特征参数,并以线性拟合度R2≥0.99、热斑电池的发热功率PS≥50 W作为热斑检测的判定条件;基于光伏组件热斑原理,以及热斑电池漏电流和反偏电压特征,分析了热斑组件的无阶梯型I-V曲线和有阶梯型I-V曲线两种I-V曲线类型。     

积灰阴影遮挡对光伏发电系统的影响

阐述了太阳能光伏电池积灰的成因、积灰的物理及化学性质和形态分类,解释了积灰的遮挡效应、腐蚀效应和热斑效应,应用MATLAB应用软件搭建光伏电池及光伏发电仿真系统,研究积灰阴影遮挡对光伏发电系统的影响。分析光伏发电系统的原理,利用MATLAB建立光伏发电系统模型,利用搭建好的模型针对光伏组件的积灰与局部阴影来进行仿真,用定性和定量分析积灰与阴影遮挡对光伏组件及其发电系统的影响,指出了对于光伏板组件上积灰清洗的重要性。     

光伏组件热斑案例失效分析与影响因素研究

通过调研实际运行光伏电站中光伏组件热斑失效数据,定义典型热斑类别,分析电池间显著温差、电池串失效、玻璃与电池碎裂等不同热斑类型的产生原因与机理。在此基础上选择不同漏电流与缺陷类型的太阳电池,封装成特定组件,实验测试并分析电池漏电流随温度的变化趋势,以及漏电流分布与电池热斑温度的相关性。并设计光伏系统模拟太阳电池失配情况,进行热斑试验,验证热斑对组件电性能输出的影响,并分析最严重热斑产生条件。     

基于注意力机制的光伏热斑识别

为解决光伏的红外热图像含有大量噪声且不同状态红外图像分布不均衡导致的热斑难以识别的问题,以Vision Transformer(ViT)模型为基础,利用卷积神经网络改进模型特征提取,利用紧凑多头自注意力机制改进模型结构,提出一种光伏红外图像热斑识别模型ConCViT,利用CIFAR-10数据集对注意力权值进行预训练,以低信噪比小样本光伏红外图像为数据集,训练出高准确率的热斑检测模型。实验结果表明,ConCViT模型比传统卷积神经网络的识别准确率高12.02%,比深度卷积自编码网络的识别准确率高4.14%,并具有更快的收敛速度。     

基于深度迁移学习的小样本光伏热斑识别方法

针对训练传统深度学习模型需大量数据而热斑效应样本数量相对少且不易采集的问题,提出基于深度迁移学习的小样本光伏热斑识别方法.在Inception-v3模型的基础上构建深度迁移学习模型,然后在负样本多分类的小样本热斑数据集上完成训练,得到可用于热斑识别的网络模型.实验结果表明,在样本数量不充足的情况下深度迁移学习方法训练出...     

基于可见光和红外热图像融合的光伏阵列热斑检测方法

利用红外热图像来检测光伏阵列的热斑时,由于复杂环境的干扰(例如阳光反射),红外热图像会显示错误的热斑而导致虚警,传统的图像处理方法对此效果不佳。该文提出一种基于可见光和红外热图像融合的热斑检测方法,能消除错误的热斑虚警,提高检测热斑的准确率。实验结果表明:该方法对复杂环境干扰下的运行中光伏阵列较有意义,从而高效便捷地检测出热斑故障。     

基于可见光和红外热图像融合的光伏阵列热斑检测方法

利用红外热图像来检测光伏阵列的热斑时,由于复杂环境的干扰(例如阳光反射),红外热图像会显示错误的热斑而导致虚警,传统的图像处理方法对此效果不佳。该文提出一种基于可见光和红外热图像融合的热斑检测方法,能消除错误的热斑虚警,提高检测热斑的准确率。实验结果表明:该方法对复杂环境干扰下的运行中光伏阵列较有意义,从而高效便捷地检测出热斑故障。     

基于计算流体力学的数据中心机器学习热模型数据增强技术

基于人工智能的热模型可以有效地提升数据中心制冷能效比.受到机房实际采集数据的数量不足和覆盖范围不足的影响,使用实际采集数据集训练的热模型常常在准确度和泛化能力上存在不足.文章介绍了一种基于计算流体力学(Computational Fluid Dynamics,CFD)的人工合成数据增强技术,采用增强数据源对人工智能热模...     

TRANSIENT THERMOELASTIC ANALYSIS OF DISK BRAKE USING THE FAST FOURIER TRANSFORM AND FINITE ELEMENT METHOD

This article introduces a finite element method that combines the fast Fourier transform technique and a conventional finite element method as a computational technique for investigating a thermomechanical problem. The conventional finite element formulation is very inefficient in the analysis of a three-dimensional disk brake model of a rotating axisymmetric disk subjected to a nonaxisymmetric transient heat flux condition due to frictional contact with asymmetric pads fixed in space. Because the proposed technique reduces the three-dimensional disk brake mathematics to two dimensions, is an extreme time saver, and costs less, we can solve the transient thermoelastic problem and the thermoelastic instability. As a result of the study we present some analyses on temperature distributions and displacement distributions in a disk brake system at a low speed and on the hot spots at a high speed above critical speed.     

Inverse problem of estimating space and time dependent hot surface heat flux in transient transpiration cooling process

In this work, an inverse problem of the transient transpiration cooling is investigated in detail. The hot surface heat flux, which is dependent on time and space, is estimated according to the temperatures measured by thermal sensors. The local thermal non-equilibrium (LTNE) model is used to describe the energy conservation of transpiration cooling process and the thermal dispersion of coolant fluid in considered. The conjugate gradient method (CGM) is applied to solve this inverse problem. The accuracy of the inverse solutions is examined by a certain heat flux with given measurement errors. The examination shows that the inverse method presented by this work can obtain satisfactory results. The effect of variable thermal properties on the inverse solutions cannot be neglected since the large temperature gradient close to the hot surface. The suitable measurement times and points should be chose by considering acceptable accuracy, computational time and memory. Meanwhile, it is found out that the location should be close to the hot surface for more accuracy.     

基于改进Mask-RCNN模型的电力设备锈蚀识别

针对变电站电力设备锈蚀检测问题,文章提出了一种基于改进Mask-RCNN模型的锈蚀检测方法.使用残差网络Resnet101作为模型的基础网络,使用改进后的非极大值抑制(Non-Maximum Suppression,NMS)算法,提升语义分割的检测准确率.针对设备锈蚀数据集,使用dropout算法对Mask-RCNN提...     

Numerical modeling of shock-to-detonation transition in energetic materials

Determining the hazard classification of energetic materials is important for transportation safety and storage concerns. To avoid costly grain redesign and additional testing, a model that adequately predicts the shock sensitivity of energetic materials is required, particularly the outcome of the Naval Ordnance Laboratory Large Scale Gap Test. The goals of this effort are to develop and validate computational tools that predict the shock sensitivity of energetic materials. Specifically, to use our packing code, Rocpack, to generate morphologies of interest for shock sensitivity assessments, and to use our CFD code, RocSDT, to propagate shocks of various strengths through the pack to predict the onset of detonation.Dealing accurately with the material interfaces in this problem is a long-standing challenge, as familiar strategies lead to spurious temperature spikes, and therefore spurious reaction rate spikes. We describe a new strategy, which does not generate spurious spikes, and demonstrate via a number of test problems that numerical convergence can be achieved. We also examine two problems that are stepping stones to a complete simulation; both are planar. In the first, we consider the passage of a shock wave through pure HMX in which a line of hot spots of the kind generated by void collapse are located a short distance behind the shock. When the hot spot spacing is large, the shock remains a shock; when small, transition to detonation occurs. In the second problem we also insert hot spots, but into a matrix of HMX particles and binder.     

A thermohydraulic modelling for the stator bars of largeturbogenerators: development, validation by laboratory and on site tests

This paper presents a thermohydraulic model of stator bars in large turbogenerators. From the losses dissipated in a bar and considering the heat transfer by conduction and convection, it calculates the hot spot temperatures in it and the slot thermocouple temperature, for a clean or polluted stator. This model has been validated by laboratory tests made on individual bars and also by on site measurements made on a 900 MW turbogenerator. It shows the difficulty for the detection of hot spots by a slot thermocouple, and its systematic detection by the water temperature measurement at the bar outlet end. The modelling may be applicable for water-cooled machines ranging from 600 MW up to 1500 MW