Automatic diagnosis of diabetic retinopathy using vision transformer based on wide-field optical coherence tomography angiography

Diabetic retinopathy(DR)is one of the major causes of visual impairment in adults with diabetes.Optical coherence tomography angiography(OCTA)is nowadays widely used as the golden criterion for diagnosing DR.Recently,wide-field OCTA(WF-OCTA)provided more abundant information including that of the peripheral retinal degenerative changes and it can contribute in accurately diagnosing DR.The need for an automatic DR diagnostic system based on WF-OCTA pictures attracts more and more attention due to the large diabetic population and the prevalence of retinopathy cases.In this study,automatic diagnosis of DR using vision transformer was performed using WF-OCTA images(12 mm×12 mm single-scan)centered on the fovea as the dataset.WF-OCTA images were automatically classified into four classes:No DR,mild nonproliferative diabetic retinopathy(NPDR),moderate to severe NPDR,and proliferative diabetic retinopathy(PDR).The proposed method for detecting DR on the test set achieves accuracy of 99.55%,sensitivity of 99.49%,and specificity of 99.57%.The accuracy of the method for DR staging reaches up to 99.20%,which has been proven to be higher than that attained by classical convolutional neural network models.Results show that the automatic diagnosis of DR based on vision transformer and WF-OCTA pictures is more effective for detecting and staging DR.

基于残差U-Net和自注意力Transformer编码器的磁场预测方法

利用有限元方法对几何结构复杂的电机和变压器进行磁场分析,存在仿真时间长且无法复用的问题。因此,该文提出一种基于残差U-Net和自注意力Transformer编码器的磁场预测方法。首先建立永磁同步电机(PMSM)和非晶合金变压器(AMT)有限元模型,得到深度学习训练所需的数据集;然后将Transformer模块与U-Net模型结合,并引入短残差机制建立ResUnet-Transformer模型,通过预测图像的像素实现磁场预测;最后通过Targeted Dropout算法和动态学习率调整策略对模型进行优化,解决拟合问题并提高预测精度。计算实例证明,ResUnet-Transformer模型在PMSM和AMT数据集上测试集的平均绝对百分比误差(MAPE)均小于1%,且仅需500组样本。该文提出的磁场预测方法能减少实际工况和多工况下精细模拟和拓扑优化的时间和资源消耗,亦是虚拟传感器乃至数字孪生的关键实现方法之一。

基于Transformer的多尺度遥感语义分割网络

为了提升遥感图像语义分割效果,本文针对分割目标类间方差小、类内方差大的特点,从全局上下文信息和多尺度语义特征2个关键点提出一种基于Transformer的多尺度遥感语义分割网络(muliti-scale Transformer network,MSTNet)。其由编码器和解码器2个部分组成,编码器包含基于Transformer改进的视觉注意网络(visual attention network,VAN)主干和基于空洞空间金字塔池化(atrous spatial pyramid pooling, ASPP)结构改进的多尺度语义特征提取模块(multi-scale semantic feature extraction module, MSFEM)。解码器采用轻量级多层感知器(multi-layer perception,MLP)配合编码器设计,充分分析所提取的包含全局上下文信息和多尺度表示的语义特征。MSTNet在2个高分辨率遥感语义分割数据集ISPRS Potsdam和LoveDA上进行验证,平均交并比(mIoU)分别达到79.50%和54.12%,平均F1-score(m F1)分别达到87.46%和69.34%,实验结果验证了本文所提方法有效提升了遥感图像语义分割的效果。

基于MTF-Swin Transformer的风机齿轮箱故障诊断

针对风机齿轮箱实际工况复杂多变及含有强噪声,传统故障诊断方法对风机齿轮箱故障诊断识别准确率较低的问题,文章提出了MTF-Swin Transformer风机齿轮箱故障诊断模型。首先,采用马尔科夫变迁场(MTF)图形编码方法将原始一维振动时序信号转化为具有关联时间信息的二维特征图谱;然后,将特征图谱作为Swin Transformer模型的输入,基于自注意力机制进行自动特征提取;最后,实现对不同故障类型的分类。仿真结果表明,该方法对齿轮箱故障诊断准确率达到了99.48%,证明了该方法的有效性和优越性。

基于CNN与Transformer相融合的心跳分类算法

心电图(Electrocardiogram,ECG)对于心血管疾病的诊断有着重要的作用,精确的心跳分类有助于后续的疾病治疗。但原始的心电图信号存在的大量噪声会影响计算机的判断,而去噪可能会导致信号特征丢失。为此,提出了利用格拉姆角场(Gramian Angular Field,GAF)将一维信号转化为图像,以保证信息完整性,然后利用计算机视觉处理技术分析心跳信号。此外,针对视觉Transformer(Vision Transformer,ViT)在局部特征捕获能力受限的问题,提出了在ViT模型内部和外部引入带残差连接的卷积神经网络(Convolutional Neural Network,CNN),使整体网络能够同时关注图像的局部特征和全局特征。同时,通过将ViT中单一的自注意块改进为双支路通道空间注意力结构,提升了模型的整体分类性能。实验结果表明,所提出的方法有效提高了网络的性能。

基于FPGA的卷积神经网络和视觉Transformer通用加速器

针对计算机视觉领域中基于现场可编程逻辑门阵列(FPGA)的传统卷积神经网(CNN)络加速器不适配视觉Transformer网络的问题,该文提出一种面向卷积神经网络和Transformer的通用FPGA加速器。首先,根据卷积和注意力机制的计算特征,提出一种面向FPGA的通用计算映射方法;其次,提出一种非线性与归一化加速单元,为计算机视觉神经网络模型中的多种非线性和归一化操作提供加速支持;然后,在Xilinx XCVU37P FPGA上实现了加速器设计。实验结果表明,所提出的非线性与归一化加速单元在提高吞吐量的同时仅造成很小的精度损失,ResNet-50和ViT-B/16在所提FPGA加速器上的性能分别达到了589.94 GOPS和564.76 GOPS。与GPU实现相比,能效比分别提高了5.19倍和7.17倍;与其他基于FPGA的大规模加速器设计相比,能效比有明显提高,同时计算效率较对比FPGA加速器提高了8.02%~177.53%。

基于局部Transformer的泰语分词和词性标注联合模型

泰语分词和词性标注任务二者之间存在高关联性,已有研究表明将分词和词性标注任务进行联合学习可以有效提升模型性能,为此,提出了一种针对泰语拼写和构词特点的分词和词性标注联合模型。针对泰语中字符构成音节,音节组成词语的特点,采用局部Transformer网络从音节序列中学习分词特征;考虑到词根和词缀等音节与词性的关联,将用于分词的音节特征融入词语序列特征,缓解未知词的词性标注特征缺失问题。在此基础上,模型采用线性分类层预测分词标签,采用线性条件随机场建模词性序列的依赖关系。在泰语数据集LST20上的试验结果表明,模型分词F1、词性标注微平均F1和宏平均F1分别达到96.33%、97.06%和85.98%,相较基线模型分别提升了0.33%、0.44%和0.12%。

Research on buffer structure and flow field simulation of swash plate plunger type hydraulic transformer

In order to solve the problem of excessive noise and vibration during the operation of the hydraulic transformer,an optimization method of valve plate damping hole structure is proposed to alleviate the phenomenon of pressure shock.Firstly,the mathematical model of oil pressure gradient in the plunger cavity is established,and the incremental equation of pressure change is derived.Secondly,a kind of buffering structure is proposed,the corresponding relationship between the pressure change and the envelopment angle of the buffering hole and the aperture size is determined by analyzing the oil pressure change curve in the plunger cavity.Finally,the flow field models with buffering holes are established,and the transient simulation of the pressure change process under the optimal solution is carried out with ANSYS software and the flow field pressure distribution contours are obtained.Through the analysis of simulation results,it is concluded that the optimal envelope angle of the three buffer holes ofA-T-B-Ais 5°,and the optimal aperture is 1.8 mm,1.6 mm,and 1.7 mm,respectively.The buffer hole can achieve a better-buffering effect in the range of variable pressure angle[0°,101°].The buffer hole structure can effectively alleviate the pressure shock and reduce the noise level,which lays a foundation for the design and theoretical research of hydraulic transformers.

Simultaneous resonance of an axially moving ferromagnetic thin plate under a line load in a time-varying magnetic field

In this paper,the simultaneous resonance of a ferromagnetic thin plate in a time-varying magnetic field,having axial speed and being subjected to a periodic line load,is studied.Based on the large deflection theory of thin plates and electromagnetic field theory,the nonlinear vibration differential equation of the plate is obtained by using the Hamilton′s principle and the Galerkin method.Then the boundary condition in which the longer opposite sides are clamped and hinged is considered.The dimensionless nonlinear differential equations are solved by using the method of multiple scales,and the analytical solution is given.In addition,the stability analysis is also carried out by using Lyapunov stability theory.Through numerical analysis,the variation curves of system resonance amplitude with frequency tuning parameter,magnetic field strength and external excitation amplitude are obtained.Different parameters that have significant effects on the response of the system,such as the thickness,the axial velocity,the magnetic field intensity,the position,and the frequency of external excitation,are considered and analyzed.The results show that the system has multiple solution regions and obvious nonlinear coupled characteristics.

Semi-analytical investigation of heat transfer in a porous convective radiative moving longitudinal fin exposed to magnetic field in the presence of a shape-dependent trihybrid nanofluid

The thermal examination of a non-integer-ordered mobile fin with a magnetism in the presence of a trihybrid nanofluid(Fe_3O_4-Au-Zn-blood) is carried out. Three types of nanoparticles, each having a different shape, are considered. These shapes include spherical(Fe_3O_4), cylindrical(Au), and platelet(Zn) configurations. The combination approach is utilized to evaluate the physical and thermal characteristics of the trihybrid and hybrid nanofluids, excluding the thermal conductivity and dynamic viscosity. These two properties are inferred by means of the interpolation method based on the volume fraction of nanoparticles. The governing equation is transformed into a dimensionless form, and the Adomian decomposition Sumudu transform method(ADSTM) is adopted to solve the conundrum of a moving fin immersed in a trihybrid nanofluid. The obtained results agree well with those numerical simulation results, indicating that this research is reliable. The influence of diverse factors on the thermal overview for varying noninteger values of γ is analyzed and presented in graphical representations. Furthermore, the fluctuations in the heat transfer concerning the pertinent parameters are studied. The results show that the heat flux in the presence of the combination of spherical, cylindrical, and platelet nanoparticles is higher than that in the presence of the combination of only spherical and cylindrical nanoparticles. The temperature at the fin tip increases by 0.705 759% when the value of the Peclet number increases by 400%, while decreases by 11.825 13% when the value of the Hartman number increases by 400%.

Analysis of Non-Sinusoidal Steady Electric Field of ±500 kV Converter Transformer

The line side winding is under the fundamental frequency AC voltage, while the valve side winding contains not only fundamental AC voltage component, but also the DC voltage component, fundamental AC voltage component, and higher harmonic voltage components when the converter transformer is at its normal operating condition, and the electric field of converter transformer is a non-sinusoidal steady one. To analyze the non-sinusoidal steady electric field containing the DC component, fundamental AC and higher harmonic components, the voltage spectrum of the valve winding in a ±500 kV converter transformer is firstly analyzed, and the non-sinusoidal periodic steady electric field is obtained by the fast discrete Fourier transform. Different resistivity of the oil and oil-immersed paper is adopted to simulate the aging of oil paper insulation at operation, and get the non-sinusoidal steady electric field.

基于改进Swin Transformer的滑坡分割算法

随着遥感技术的逐步成熟,滑坡目标分割技术在灾害防控、城市建设等领域发挥着越来越关键的作用。然而,面临着卷积神经网络在理解上下文语义信息上存在困难以及滑坡数据集中正负样本分布不均的问题,需要寻找一种更优的解决方案。为此,提出了一种改进Swin Transformer的滑坡分割算法。首先,将Swin Transformer引入到滑坡目标分割任务中,并构建了Swin-Unet网络。其次,为了提高网络的表征能力,在Swin-Unet中加入通道注意力模块;为了平衡小样本正负样本不均衡,使用在线困难样例挖掘;为了进一步提高模型性能,引入并优化了条件随机场。实验结果表明,相较于基线算法,改进方法在滑坡目标分割任务上取得了更优的性能,Dice提高了1.9%,交并比(IoU)提高了1.2%,研究为滑坡目标分割提供了一种有效的方法,并为相关领域的研究提供了新的思路。

Self-Adaptive and Electric Field-Driven Protective Layer with Anchored Lithium Deposition Enable Stable Lithium Metal Anode

Lithium metal battery has great development potential because of its lowest electrochemical potential and highest theoretical capacity.However,the uneven deposition of Li^(+)flux in the process of deposition and stripping induces the vigorous growth of lithium dendrites,which results in severely battery performance degradation and serious safety hazards.Here,the tetragonal BaTiO3 polarized by high voltage corona was used to build an artificial protective layer with uniform positive polarization direction,which enables uniform Li^(+)flux.In contrast to traditional strategies of using protective layer,which can guide the uniform deposition of lithium metal.The ferroelectric protective layer can accurately anchor the Li^(+)and achieve bottom deposition of lithium due to the automatic adjustment of the electric field.Simultaneously,the huge volume changes caused by Li^(+)migration change of the lithium metal anode during charging and discharging is functioned to excite the piezoelectric effect of the protective layer,and achieve seamless dynamic tuning of lithium deposition/stripping.This dynamic effect can accurately anchor and capture Li^(+).Finally,the layer-modified Li anode enables reversible Li plating/stripping over 1500 h at 1 mA cm^(-2)and 50℃in symmetric cells.In addition,the assembled Li-S full cell exhibits over 300 cycles with N/P≈1.35.This work provides a new perspective on the uniform Li^(+)flux at the Li-anode interface of the artificial protective layer.

Phase field model for electric-thermal coupled discharge breakdown of polyimide nanocomposites under high frequency electrical stress

In contrast to conventional transformers, power electronic transformers, as an integral component of new energy power system, are often subjected to high-frequency and transient electrical stresses, leading to heightened concerns regarding insulation failures. Meanwhile, the underlying mechanism behind discharge breakdown failure and nanofiller enhancement under high-frequency electrical stress remains unclear. An electric-thermal coupled discharge breakdown phase field model was constructed to study the evolution of the breakdown path in polyimide nanocomposite insulation subjected to high-frequency stress. The investigation focused on analyzing the effect of various factors, including frequency, temperature, and nanofiller shape, on the breakdown path of Polyimide(PI) composites. Additionally, it elucidated the enhancement mechanism of nano-modified composite insulation at the mesoscopic scale. The results indicated that with increasing frequency and temperature, the discharge breakdown path demonstrates accelerated development, accompanied by a gradual dominance of Joule heat energy. This enhancement is attributed to the dispersed electric field distribution and the hindering effect of the nanosheets. The research findings offer a theoretical foundation and methodological framework to inform the optimal design and performance management of new insulating materials utilized in high-frequency power equipment.

Research on Numerical Simulation of 3D Leakage Magnetic Field and Short-circuit Impedance of Axial Dual-low-voltage Split-winding Transformer

It is difficult to accurately calculate the short-circuit impedance, due to the complexity of axial dual-low-voltage split-winding transformer winding structure. In this paper, firstly, the leakage magnetic field and short-circuit impedance model of axial dual-low-voltage split-winding transformer is established, and then the 2D and 3D leakage magnetic field are analyzed. Secondly, the short-circuit impedance and split parallel branch current distribution in different working conditions are calculated, which is based on field-circuit coupled method. At last, effectiveness and feasibility of the proposed model is verified by comparison between experiment, analysis and simulation. The results showed that the 3D analysis method is a better approach to calculate the short-circuit impedance, since its analytical value is more closer to the experimental value compared with the 2D analysis results, the finite element method calculation error is less than 2%, while the leakage flux method maximum error is 7.2%.

Equivalent Doping Transformation Method for Predicting Breakdown Voltage and Peak Field at Breakdown of Epitaxial-Diffused Punch-Through Junction

Based on a new semi empirical analytical method, namely equivalent doping transformation, the breakdown voltage and the peak field of the epitaxial diffused punch through junction have been obtained. The basic principle of this method is introduced and a set of breakdown voltage and peak field plots are provided for the optimum design of the low voltage power devices. It shows that the analytical results coincide with the previous numerical simulation well.

Improvement of Electrical Performance of SOI-LIGBT by Resistive Field Plate

The electrical performance including breakdown voltage and turn-off speed of SOI-LIGBT is improved by incorporating a resistive field plate (RFP) and a p-MOSFET.The p-MOSFET is controlled by a signal detected from a point of the RFP.During the turning-off of the IGBT,the p-MOSFET is turned on,which provides a channel for the excessive carriers to flow out of the drift region and prevents the carriers from being injected into the drift region.At the same time,the electric field affected by the RFP makes the excessive carriers flow through a wider region,which almost eliminates the second phase of the turning-off of the SOI-LIGBT caused by the substrate bias.Faster turn-off speed is achieved by above two factors.During the on state of the IGBT,the p-MOSFET is off,which leads to an on-state performance like normal one.At least,the increase of the breakdown voltage for 25% and the decrease of the turn-off time for 65% can be achieved by this structure as can be verified by the numerical simulation results.

COUPLED NUMERICAL SIMULATION ON COLD ROLLER'S TEMPERATURE FIFLD-PHASE TRANSFORMATION - STRESS FIELD DURING ITS QUENCHING PSOCESS

Complicated changns occur inside the steel parts during quenching process. The abruptly changed boundary conditions make the temperature field,micro - structure and stress field change dramatically in very short time, and these variables take a contact interactions in the whole process. In this paper, a three dimensional non - linear mathematical model for queeching process has been founded and the numerical simulation on temperature field,microstructre and stress field has been realized.In the FEM analysis, the incremental iteration method is used to deal with such complicated nonlinear as boundary nonlinear, physical property nonlinear,transformation nonlinear etc.The effect of stress on transformation kinetics has been considered in the calculation of microstructure. In the stress field anal- ysis,a thermo- elasto - plastic model has been founded, which considers such factors as transforma- tion strain,transformation plastic strain, themal strain and the effect of temperature and transforma- tion on mechanical propertier etc. The transient temperature field, microstructure distribution and stress field of the roller on any time can be displayed vividly,and the cooling curve and the changes of stress on any position can also be given.

Optical Sensor with Wide Range and High Sensitivity for Internal Magnetic Field Detection of Transformer

The deterioration of winding defects is one of the important causes of power transformer fires and even explosion failures.The change of leakage magnetic field distribution is the most direct response to winding defects.Currently there are few sensors suitable for online measurement of the internal magnetic field of transformers.Based on the Faraday magneto-optical effect,a magnetic field sensor with wide range and high sensitivity is proposed in this paper,which is suitable for the interior use of transformers.The straight-through optical structure with interior polarizer is adopted,and the sensor has a measurement range of 1.5 T and a sensitivity of 1 mT.It also possesses a small size,with a length of about 30 mm after encapsulation.The influence mechanism of vibration and temperature is revealed through theoretical analysis and numerical simulation.It is proposed to filter out the interference of vibration by characteristic frequency analysis and to compensate for temperature by a two-probe structure.An anti-interference test verifies the effectiveness of this method,and it can reduce the error from 80.56%to 2.63%under the combined interference of vibration and temperature.

Magnetoelastic combined resonance and stability analysis of a ferromagnetic circular plate in alternating magnetic field

The nonlinear combined resonance problem of a ferromagnetic circular plate in a transverse alternating magnetic field is investigated. On the basis of the deformation potential energy, the strain potential energy, and the kinetic energy of the circular plate, the Hamilton principle is used to induce the magnetoelastic coupling transverse vibration dynamical equation of the ferromagnetic circular plate. Based on the basic electromagnetic theory, the expressions of the magnet force and the Lorenz force of the circular plate are presented. A displacement function satisfying clamped-edge combined with the Galerkin method is used to derive the Duffing vibration differential equation of the circular plate. The amplitude-frequency response equations of the system under various combined resonance forms are obtained by means of the multi-scale method, and the stability of the steady-state solutions is analyzed according to the Lyapunov theory. Through examples, the amplitude-frequency characteristic curves with different parameters, the amplitude of resonance varying with magnetic field intensity and excitation force, and the time-course response diagram, phase diagram, Poincar′e diagram of the system vibration are plotted, respectively. The effects of different parameters on the amplitude and stability of the system are discussed. The results show that the electromagnetic parameters have a significant effect on the multi-valued attribute and stability of the resonance solutions, and the system may exhibit complex nonlinear dynamical behavior including multi-period and quasi-periodic motion.