High Frequency Transformers for DC/DC Converter Used in Solar PV System

This paper presents and investigates planar and coaxial high frequency power transformers used for DC/DC converters in a three phase photo voltaic （PV） power systems. The winding structure including a Faraday shield between the primary and secondary windings is designed to minimize eddy current losses, skin and proximity effects, and to reduce the leakage inductance, and the inter winding coupling capacitance. Finite Element Method is employed to analyze the magnetic flux and eddy current distributions. The two different kinds of prototype high frequency transformers are designed and tested. The simulation and experiment results are demonstrated and compared with non-shielded transformers. The shielded transformers have achieved the expected results with a relatively small coupling capacitance, compared with the conventional high frequency transformer. This shield decreases the inter-winding coupling capacitance Cps. The topology of this shield has to be such that it acts as a Faraday screen while avoiding eddy current generation.

基于边缘引导和动态可变形Transformer的遥感图像变化检测

卷积神经网络(Convolutional Neural Network,CNN)和Transformer的混合架构能够有效建模图像的局部与全局特征,已成为遥感图像变化检测任务的主流网络.然而这类网络仍面临着一些挑战. CNN分支中的卷积和池化运算通常会抑制遥感图像中的高频信息,降低目标边界的精度;此外,Transformer分支对图像像素进行等同长程依赖关系建模,忽略了变化目标的形状及语义关联信息,导致网络对变化目标特征的表达不足.为解决上述问题,提出了基于边缘引导和动态可变形Transformer的遥感图像变化检测网络.在CNN分支中设计了边缘信息引导模块,利用高频信息增强目标区域的边缘信息,从而改善变化目标的轮廓精度.同时设计了一种新颖的动态可变形Transformer,能够自适应地匹配形状不同的变化目标,选择与变化相关的特征建模长程依赖关系,以提高网络的特征表达能力.实验结果表明,提出的方法在三个公开数据集LEVIR-CD、CDD和DSIFN-CD上显著提高了检测精度,在变化目标的边界精度和内部完整性方面都明显优于当前的主流网络.

Novel forward mode AC/AC converters with high frequency link

A circuit configuration and a circuit topologic family of the novel forward mode AC/AC converters with high frequency link are presented. The circuit configuration is constituted of input cycloconverter, high frequency transformer, output cycloconverter, input and output filters. The circuit topologic family includes eight circuit topologies, such as full-bridge-full-wave mode, etc. The bi-polarity phase-shifted control strategy and steady principles are thoroughly investigated. The output characteristics are obtained. By using the bi-polarity phase-shifted control strategy with phase-shifted control between the output cycloconveter and the input cycloconverter, commutation overlap period of the output cycloconverter, and polarity selection of the output filtering inductance current and the input voltage, the leakage inductance energy and the output filtering inductance current are naturally commutated, and surge voltage and surge current of the cycloconverters are overcome. The converters have such advantages as simple topology, two-stage power conversions(LFAC/HFAC/LFAC), bi-directional power flow, high frequency electrical isolation, good output waveforms, and strong ability to stabilize voltage. The converters lay key technical foundation on a new-type of regulated sinusoidal AC power supplies and electronic transformers. The correction and advancement of the converters are well verified by a principle test.

小波分频自注意力Transformer图像去雨网络

针对视觉Transformer对高频信息捕捉能力弱以及目前许多图像去雨方法易丢失细节的问题,提出小波分频自注意力Transformer图像去雨网络(WFDST-Net)。小波分频自注意力Transformer(WFDST)作为WFDST-Net的主要模块,其利用不可分提升小波变换获取特征图的低频分量和高频分量,分别在低频和高频中进行自注意力交互,使模块从低频中学习恢复全局结构的能力,在高频中强化捕捉雨纹等线条细节的能力,增强对不同频域特征的建模能力。WFDST-Net采用U形架构并通过不可分提升小波变换获取多尺度特征,可在捕获不同形状高频雨纹的同时保证信息的完整性。相比其他图像去雨相关的Transformer,WFDST-Net具有更低的参数量。此外,提出VOCRain250数据集用于联合图像去雨和语义分割任务,该数据集比目前广泛使用的BDD150更具优势。实验表明,所提方法增强了视觉Transformer对不同频域信息的捕获能力,并在合成和真实数据集以及VOCRain250中的表现优于目前先进的去雨方法,能有效去除复杂雨纹并保留更多细节特征。

基于DFT的频率敏感双分支Transformer多变量长时间序列预测方法

在进行多变量长时间序列预测时,仅利用时域分析通常无法充分捕捉长时间序列依赖,而这会导致信息利用率不足、预测精度不够高。因此,结合频域时域分析,提出一种基于离散傅里叶变换(DFT)的频率敏感双分支多变量长时间序列预测(FSDformer)方法。首先,通过DFT实现时间和频率的相互转换,从而将复杂的时间序列数据分解为结构简单的低频趋势项、中频季节项和高频余项3个分量;其次,采用双分支结构,针对中高频分量预测,应用Encoder-Decoder结构,设计了周期性增强注意力机制;针对低频趋势分量预测,采用多层感知机(MLP)结构;最后将中高频分量与低频分量预测结果相加,得到多变量长时间序列的最终预测结果。在2个数据集上把FSDformer与其他5个经典算法进行了对比分析,在Electricity数据集上,当历史序列长度为96,预测序列长度为336时,相较于Autoformer等对比算法,FSDformer的平均绝对误差(MAE)下降了11.5%~29.1%,均方误差(MSE)下降了20.9%~43.7%,达到了最优预测精度。实验结果表明,FSDformer能有效捕捉长时间序列的相关依赖,在提升预测精度和计算效率的同时,增强了模型预测的稳定性。

Exploration of instantaneous frequency for local control assessment in real-time hybrid simulation

Local control parameters such as instantaneous delay and instantaneous amplitude play an essential role in evaluating the performance and maintaining the stability of real-time hybrid simulation(RTHS).However,existing methods have limitations in obtaining this local assessment in either the time domain or frequency domain.In this study,the instantaneous frequency is introduced to determine local control parameters for actuator tracking assessment in a real-time hybrid simulation.Instantaneous properties,including amplitude,delay,frequency and phase,are then calculated based on analytic signals translated from actuator tracking signals through the Hilbert transform.Potential issues are discussed and solutions are proposed for calculation of local control parameters.Numerical simulations are first conducted for sinusoidal and chirp signals with time varying amplitude error and delay to demonstrate the potential of the proposed method.Laboratory tests also are conducted for a predefined random signal as well as the RTHS of a single degree of freedom structure with a self-centering viscous damper to experimentally verify the effectiveness of the proposed use of the instantaneous frequency.Results from the ensuing analysis clearly demonstrate that the instantaneous frequency provides great potential for local control assessment,and the proposed method enables local tracking parameters with good accuracy.

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.

基于Transformer大规模MIMO的CSI反馈网络

针对频分双工大规模多输入多输出系统中信道状态信息反馈方法反馈精度低的问题,提出了一种名为CDTransformer的基于Transformer和卷积分解(convolutional decomposition,CD)的CSI反馈网络。CDTransformer将卷积分解融入改进的Transformer结构网络中,在不增加计算复杂度的情况下提升网络性能,并通过对网络的全连接层进行二值化实现轻量级部署。针对用户端功率有限的情况,提出了一种MixedTransformer网络模型。其中编码器采用计算成本较低、结构简单的单层卷积神经网络,而解码器则采用与CDTransformer模型相同的结构。CDTransformer模型融合了Transformer结构和卷积分解,提高了CSI反馈精度并实现轻量级部署。此外,引入了MixedTransformer模型,结合了CDTransformer和卷积神经网络的优点,以在功率有限情况下提供更好的性能。结果显示,相比于CsiTransformer网络模型,CDTransformer网络模型在归一化均方误差和余弦相似度方面分别提高了37.7%和0.2%。

基于CNN-Swin Transformer Network的LPI雷达信号识别

针对在低信噪比(SNR)条件下,低截获概率雷达信号调制方式识别准确率低的问题,提出一种基于Transformer和卷积神经网络(CNN)的雷达信号识别方法。首先,引入Swin Transformer模型并在模型前端设计CNN特征提取层构建了CNN+Swin Transformer网络(CSTN),然后利用时频分析获取雷达信号的时频特征,对图像进行预处理后输入CSTN模型进行训练,由网络的底部到顶部不断提取图像更丰富的语义信息,最后通过Softmax分类器对六类不同调制方式信号进行分类识别。仿真实验表明:在SNR为-18 dB时,该方法对六类典型雷达信号的平均识别率达到了94.26%,证明了所提方法的可行性。

A high frequency vibration compensation approach for ultrahigh resolution SAR imaging based on sinusoidal frequency modulation Fourier-Bessel transform

Ultrahigh resolution synthetic aperture radar(SAR)imaging for ship targets is significant in SAR imaging,but it suffers from high frequency vibration of the platform,which will induce defocus into SAR imaging results.In this paper,a novel compensation method based on the sinusoidal frequency modulation Fourier-Bessel transform(SFMFBT)is proposed,it can estimate the vibration errors,and the phase shift ambiguity can be avoided via extracting the time frequency ridge consequently.By constructing the corresponding compensation function and combined with the inverse SAR(ISAR)technique,well-focused imaging results can be obtained.The simulation imaging results of ship targets demonstrate the validity of the proposed approach.

Enhanced Wideband Frequency Estimation via FFT: Leveraging Polynomial Interpolation and Array Indexing

Accurate frequency estimation in a wideband digital receiver using the FFT algorithm encounters challenges, such as spectral leakage resulting from the FFT’s assumption of signal periodicity. High-resolution FFTs pose computational demands, and estimating non-integer multiples of frequency resolution proves exceptionally challenging. This paper introduces two novel methods for enhanced frequency precision: polynomial interpolation and array indexing, comparing their results with super-resolution and scalloping loss. Simulation results demonstrate the effectiveness of the proposed methods in contemporary radar systems, with array indexing providing the best frequency estimation despite utilizing maximum hardware resources. The paper demonstrates a trade-off between accurate frequency estimation and hardware resources when comparing polynomial interpolation and array indexing.

基于Transformer的出铝量决策算法研究与应用

在传统电解铝工业生产中,铝电解槽的生产决策通常根据工艺技术人员多年的经验制定,其中出铝量是一个具有强耦合性的重要决策变量,其决策的好坏对电解槽的生产稳定性和产铝效率具有直接且重要的影响。本文提出了一种可深度挖掘数据频率特征和特征筛选的Transformer架构模型FDisformer,优化了传统的Transformer的encode编码层,进而挖掘数据趋势变化和更深层次的特征信息,同时引入了特征蒸馏模块,确保筛选出与出铝量决策这一任务强相关的特征。FDisformer在出铝量决策方面具有更高的性能指标,该模型的建立可以为后续的铝电解槽出铝量每日决策提供参考依据。

Analysis of Thermal Behavior of High Frequency Transformers Using Finite Element Method

High frequency transformer is used in many applications among the Switch Mode Power Supply (SMPS), high voltage pulse power and etc can be mentioned. Regarding that the core of these transformers is often the ferrite core;their functions partly depend on this core characteristic. One of the characteristics of the ferrite core is thermal behavior that should be paid attention to because it affects the transformer function and causes heat generation. In this paper, a typical high frequency transformer with ferrite core is designed and simulated in ANSYS software. Temperature rise due to winding current (Joule-heat) is considered as heat generation source for thermal behavior analysis of the transformer. In this simulation, the temperature rise and heat distribution are studied and the effects of parameters such as flux density, winding loss value, using a fan to cool the winding and core and thermal conductivity are investigated.

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.

Zero-voltage switching converter absorbing parasitic parameters for super high frequency induction heating

This paper presents a novel mega-Hz-level super high frequency zero-voltage soft-switching converter for induction heating power supplies. The prominent advantage of this topology is that it can absorb both inductive and capacitive parasitic components in the converter. The switch devices operate in a zero-voltage soft-switching mode. Consequently, the high voltage and high current spikes caused by parasitic inductors or capacitors oscillation do not occur in this circuit, and the high power loss caused by high frequency switching can be greatly reduced. A large value inductor is adopted between the input capacitor and the switches, thus, this novel converter shares the benefits of both voltage-type and current-type circuits simultaneously, and there are no needs of dead time between two switches. The working principles in different modes are introduced. Results of simulation and experiments operated at around 1 MHz frequency verify the validity of parasitic components absorption and show that this convener is competent for super high frequency applications.

Development of a Medium Voltage, High Power, High Frequency Four-Port Solid State Transformer

The power and voltage levels of renewable energy resources is growing with the evolution of the power electronics and switching module technologies.For that,the need for the development of a compact and highly efficient solid-state transformer is becoming a critical task in-order to integrate the current AC grid with the new renewable energy systems.The objective of this paper is to present the design,implementation,and testing of a compact multi-port solid-state transformer for microgrid integration applications.The proposed system has a four-port transformer and four converters connected to the ports.The transformer has four windings integrated on a single common core.Thus,it can integrate different renewable energy resources and energy storage systems.Each port has a rated power of 25 kW,and the switching frequency is pushed to 50 k Hz.The ports are chosen to represent a realistic industrial microgrid model consisting of grid,energy storage system,photovoltaic system,and load.The grid port is designed to operate at 4.16 k VAC corresponding to 7.2 kV DC bus voltage,while the other three ports operate at 500 VDC.Moreover,the grid,energy storage and photovoltaic ports are active ports with dual active bridge topologies,while the load port is a passive port with full bridge rectifier one.The proposed design is first validated with simulation results,and then the proposed transformer is implemented and tested.Experimental results show that the designed system is suitable for 4.16 k VAC medium voltage grid integration.

Modeling and Control of the Modular Multilevel Converter (MMC) based Solid State Transformer (SST) with Magnetic Integration

Solid state transformer(SST)can provide more advanced functionalities compared with conventional transformer,and has great potential in smart grid application.Recently,the SST with medium frequency(MF)isolation link and magnetic integration feature has been proposed,which can reduce the system volume and thus increase the power density.However,the magnetic integration also introduces strong coupling between the line frequency(LF)and MF variables,which poses a great challenge on modeling and control issues.This paper proposes a modeling and control method for an SST with magnetic integration and mixed-frequency modulation.A mathematical model based on dual d-q references is deduced,and then a cascaded control system is designed according to the model.Parameters of the controller for the variables at one frequency are properly designed to avoid disturbance from the variables at the other frequency.The simulation and experimental results show good decoupling effect and satisfactory dynamics performance of the proposed control system.

Development of OPTO-LDR coupled timer based voltage to frequency converter

This paper describes the development of a timer based voltage to frequency converter(V FC).Timer LM555is used in astable multivibrator mode with two OPTO-LDRs(light dependent resistors)in the circuitry.The frequency of timer output waveform which is measured using a digital storage oscillator(DSO)is almost linearly proportional to the applied input voltage.Hence we obtain a linear relationship between the frequency of timer output waveform and the input voltage.Because of its quasi-digital output,the main advantages of this developed converter are linear input-output relationship,small size,easy portabilityand high cost performance.In addition,the timer output waveform can be directly interfaced with personal computer or microprocessor/microcontroller for further processing of the input voltage signal without intervening any analog-to-digital converter(ADC).

Analysis of Soft-Switched High Frequency AC-AC Converter for Induction Motor Drive

The systematic mathematical analysis of the high frequency soft-switched AC-AC converter is proposed for variable frequency induction machine load in this paper. Both PWM and square-wave modes of operation have been considered. The frequency relation and phase unbalance problem due to discrete time integral half-cycle switching has been discussed in the beginning. Then, generalized Fourier series have been derived for output voltage, output current and supply current in two modes.The analytical results help to understand tbe converter characteristics, design optimally a convertermachine system of arbitrary capacity considering the various trade-off parameters.