期刊文献+
共找到2,016篇文章
< 1 2 101 >
每页显示 20 50 100
Mechatronic Modeling and Domain Transformation of Multi-physics Systems
1
作者 Clarence W.DE SILVA 《Instrumentation》 2021年第1期14-28,共15页
The enhanced definition of Mechatronics involves the four underlying characteristics of integrated,unified,unique,and systematic approaches.In this realm,Mechatronics is not limited to electro-mechanical systems,in th... The enhanced definition of Mechatronics involves the four underlying characteristics of integrated,unified,unique,and systematic approaches.In this realm,Mechatronics is not limited to electro-mechanical systems,in the multi-physics sense,but involves other physical domains such as fluid and thermal.This paper summarizes the mechatronic approach to modeling.Linear graphs facilitate the development of state-space models of mechatronic systems,through this approach.The use of linear graphs in mechatronic modeling is outlined and an illustrative example of sound system modeling is given.Both time-domain and frequency-domain approaches are presented for the use of linear graphs.A mechatronic model of a multi-physics system may be simplified by converting all the physical domains into an equivalent single-domain system that is entirely in the output domain of the system.This approach of converting(transforming)physical domains is presented.An illustrative example of a pressure-controlled hydraulic actuator system that operates a mechanical load is given. 展开更多
关键词 Mechatronic Modeling Multi-physics Systems Integrated Unified Unique and Systematic Approach Linear Graphs Physical domain Conversion/transformation
原文传递
基于Transformer和卷积神经网络的领域自适应翻译模型优化方法研究
2
作者 李亚琪 李冬瑞 《自动化与仪器仪表》 2026年第1期273-277,共5页
传统翻译模型容易性能退化,需要依靠大量人工标注数据,且自适应能力有限。为避免这些问题,研究设计了基于Transformer和卷积神经网络的领域自适应翻译模型,通过隐式条件位置的编码器取代Transformer的固定编码器,并在前端加入复数卷积模... 传统翻译模型容易性能退化,需要依靠大量人工标注数据,且自适应能力有限。为避免这些问题,研究设计了基于Transformer和卷积神经网络的领域自适应翻译模型,通过隐式条件位置的编码器取代Transformer的固定编码器,并在前端加入复数卷积模块,以捕捉英文的词法和句法特征,再结合注意力机制增强领域特异性模式表征。实验表明,研究设计模型的损失值稳定在5.5和4.9,收敛速度较快;对艺术类与文学作品文本翻译的耗时仅为49 ms~50 ms;模型F1分数达96%;连续翻译60次时模型顿卡率仅为21%,翻译准确率达91%。结果表明研究模型有助于提升专业领域翻译精准度与鲁棒性,尤其适用于资源受限环境下的跨领域语义精准迁移,并且平衡了计算效率与性能。为医疗、法律等领域提供了低人工成本、高适应性的翻译新路径。 展开更多
关键词 英文翻译模型 领域自适应 transformER 卷积神经网络
原文传递
Performance of Wavelet-Transform-Domain Adaptive Equalizers 被引量:4
3
作者 吴炳洋 陈琦帆 程时昕 《Journal of Southeast University(English Edition)》 EI CAS 2002年第1期13-18,共6页
In this paper performances of wavelet transform domain (WTD) adaptive equalizers based on the least mean ̄square (LMS) algorithm are analyzed. The optimum Wiener solution, the condition of convergence, the minimum ... In this paper performances of wavelet transform domain (WTD) adaptive equalizers based on the least mean ̄square (LMS) algorithm are analyzed. The optimum Wiener solution, the condition of convergence, the minimum mean square error (MSE) and the steady state excess MSE of the WTD adaptive equalizer are obtained. Constant and time varying convergence factor adaptive algorithms are studied respectively. Computational complexities of WTD LMS equalizers are given. The equalizer in WTD shows much better convergence performance than that of the conventional in time domain. 展开更多
关键词 WAVELET transform domain wavelet transform domain LMS adaptive equalizer
在线阅读 下载PDF
Fault Location in Distribution System Using Convolutional Neural Network Based on Domain Transformation 被引量:7
4
作者 Yang Yu Mengshi Li +1 位作者 Tianyao Ji Q.H.Wu 《CSEE Journal of Power and Energy Systems》 SCIE CSCD 2021年第3期472-484,共13页
Distribution lines are integral parts of the modern power system,which can affect the security and stability of power supply directly.An effective power system protection scheme should be able to detect all occurring ... Distribution lines are integral parts of the modern power system,which can affect the security and stability of power supply directly.An effective power system protection scheme should be able to detect all occurring faults as soon as possible.There are two tasks in fault diagnosis.One is the fault classification,where high accuracy rates have already achieved.Thus,this paper focuses on the other task,i.e.fault location.Enlightened by Fourier transform,this paper proposes an online data-driven method,which transforms signals from time domain to image domain through signal-to-image(SIG)algorithm and then process the transformed images with framework based on convolutional neural network(CNN).On the one hand,we can extract more crucial characteristic and information from image domain.On the other hand,the CNN-based structure is much smaller than others.It needs less memory space and would be easier to be transplanted to hardware platform.Moreover,the proposed algorithm does not require synchronous devices.The numerical comparison shows that the proposed SIG-CNN fault location model achieves robust and accurate results compared with other data-driven algorithms. 展开更多
关键词 Convolutional neural network distribution lines domain transformation fault location
原文传递
Post-Mesozoic Transformation of Tectonic Domain in Southeastern China and Its Geodynamic Mechanism 被引量:12
5
作者 Wu Ganguo Zhang Da Faculty of Earth Sciences and Mineral Resources, China University of Geosciences, Beijing 100083, China Chen Bailin Institute of Geomechanics, CAGS, Beijing 100081, China Wu Jianshe Institute of Geological Survey of Fujian Pr 《Journal of Earth Science》 SCIE CAS CSCD 2000年第3期94-98,共5页
Since the Mesozoic and Cenozoic, a transformation from a Tethyan Himalayan tectonic domain into a circum Pacific tectonic domain from Indosinian to Yanshanian is indicated in this paper, resulting in conspicuous cha... Since the Mesozoic and Cenozoic, a transformation from a Tethyan Himalayan tectonic domain into a circum Pacific tectonic domain from Indosinian to Yanshanian is indicated in this paper, resulting in conspicuous changes in geophysics, tectono magmatic distribution, lithofacies and paleo geography, tectonic system in southeastern China. Tectonic analysis shows that the tectonic framework resulted from the compounding, transforming and superimposing of the two tectonic domains. The geodynamic mechanism of the transformation is mainly shown as the transverse and longitudinal heterogeneity of lithosphere, and the exchange between the crust and the mantle. 展开更多
关键词 transformation of tectonic domain geodynamic mechanism southwestern Fujian Province southeastern China.
在线阅读 下载PDF
INTERFERENCE MITIGATING BASED ON FRACTIONAL FOURIER TRANSFORM IN TRANSFORM DOMAIN COMMUNICATION SYSTEM 被引量:6
6
作者 Wang Chuandan Zhang Zhongpei Li Shaoqian 《Journal of Electronics(China)》 2007年第2期181-186,共6页
The method of FRactional Fourier Transform (FRFT) is introduced to Transform Domain Communication System (TDCS) for signal transforming in the paper after theoretical analysis. The method yields optimal Basis Function... The method of FRactional Fourier Transform (FRFT) is introduced to Transform Domain Communication System (TDCS) for signal transforming in the paper after theoretical analysis. The method yields optimal Basis Function (BF) by FRFT with optimal transform angle. The TDCS using the proposed method has wider usable spectrum, stronger robustness and better ability of anti non-stationary jamming than using usual methods, such as Fourier Transform (FT), Auto Regressive (AR), Wavelet Transform (WT), etc. The main simulation results are as follows. First, the Bit Error Rate (BER) Pb is close to theoretical bound of no jamming no matter in single tone or in linear chirp interference. Second, the interference-to-signal ratio J /E is at least 12dB more than that of Direct Spread Spectrum System (DSSS) under the same BER if the spectrum hopping-to-signal ratio is 1:20 in chirp plus hopping interfering. Third, the Eb /N 0(when estimation difference is 90% between trans- mitter and receiver) is about 3.5dB or about 0.5dB (when estimation difference is 10% between transmitter and receiver) more than that of theoretical result when no estimation difference un-der Pb=10-2. 展开更多
关键词 transform domain Communication System (TDCS) Basis Function (BF) Fourier transform (FT) FRactional Fourier transform (FRFT) Hopping jamming
在线阅读 下载PDF
基于DRSN-Transformer编码器的域自适应辐射源个体识别方法研究
7
作者 张冠杰 李艳斌 +1 位作者 畅鑫 闫红超 《河北工业科技》 2025年第4期303-313,共11页
为了使深度神经网络能够准确识别不同传输信道的辐射源个体,提出了一种基于深度残差收缩网络(deep residual shrinkage network,DRSN)融合Transformer编码器的域自适应个体识别方法。采用DRSN软阈值模块自动去掉I/Q接收信号中的噪声,利... 为了使深度神经网络能够准确识别不同传输信道的辐射源个体,提出了一种基于深度残差收缩网络(deep residual shrinkage network,DRSN)融合Transformer编码器的域自适应个体识别方法。采用DRSN软阈值模块自动去掉I/Q接收信号中的噪声,利用Transformer编码器进一步提取信号中各符号间的依赖特征,使用域自适应对抗学习方法将不同域的目标信号映射为相同分布的目标特征,使得DRSN-Transformer编码器网络模型能够准确提取与信道无关的射频指纹特征(radio frequency fingerprint,RFF),实现信道变化时目标辐射源个体的精准识别,并利用调制器畸变信号模型进行了仿真试验。结果表明:与ResNet和DRSN网络模型相比,所提DRSN-Transformer网络模型的平均识别准确率分别提升了2.98个百分点和1.65个百分点;采用域自适应对抗学习方法的DRSN-Transformer编码器网络模型能够有效降低源域和目标域信号特征分布的不一致性,与传统方法训练的DRSN-Transformer编码器网络模型相比,在信噪比为27 dB时,识别准确率提升了20.73个百分点,显著改善了信道变化时的辐射源个体识别性能。与传统学习方法相比,所提方法虽然增加了特征提取网络与域判别网络的对抗训练过程,但训练完成的特征提取网络能够准确提取与信道变化无关的指纹特征,在辐射源个体识别领域具有一定的应用价值。 展开更多
关键词 计算机神经网络 深度残差收缩网络 transformer编码器 域对抗神经网络 特定辐射源识别 信道自适应
在线阅读 下载PDF
基于自适应Transformer的短期负荷预测域适应方法 被引量:1
8
作者 周俊 马泽菊 《重庆电力高等专科学校学报》 2025年第1期1-6,共6页
有效的STLF对于当今电力系统的平稳运行至关重要。然而,传统的预测方法往往无法应对电力消费数据中的复杂性、非线性和动态变化,特别是在处理具有不同数据分布的新区域时。为了克服这些限制,提出了针对STLF的ATDA。ATDA利用Transformer... 有效的STLF对于当今电力系统的平稳运行至关重要。然而,传统的预测方法往往无法应对电力消费数据中的复杂性、非线性和动态变化,特别是在处理具有不同数据分布的新区域时。为了克服这些限制,提出了针对STLF的ATDA。ATDA利用Transformer编码器有效的建模时间依赖性,并结合带有重要性加权的部分对抗域适应策略,解决源域和目标域之间的差异。通过优先考虑与目标域最相关的源样本,ATDA最小化了负迁移并提高了预测精度。在来自国家电网公司的真实数据上进行的综合实验表明,ATDA在预测性能上显著优于当前领先的模型。 展开更多
关键词 短期负荷预测 域适应 transformER 对抗学习 重要性加权
在线阅读 下载PDF
Transformations between Aeromagnetic Gradients in Frequency Domain 被引量:3
9
作者 李海侠 徐世浙 +2 位作者 余海龙 魏巍 房江奇 《Journal of Earth Science》 SCIE CAS CSCD 2010年第1期114-122,共9页
Aeromagnetic gradients are often used to enhance details or add new insights for interpretation. The gradients may be measured or derived from the total field or from transformation between horizontal and vertical gra... Aeromagnetic gradients are often used to enhance details or add new insights for interpretation. The gradients may be measured or derived from the total field or from transformation between horizontal and vertical gradients. At present, vertical, horizontal, and triaxial aeromagnetic gradiometers are in operation throughout the world, while the first two are used more widely. Transformations between horizontal and vertical gradients are needed for acquiring three gradient components or for checking the validity of measured gradients. Transformation of potential field by fast Fourier transform technique in frequency domain is popularly used; however, when applied to transforming between gradients, there is a problem that needs resolving. Because those expressions of transform operators are undefined when u or v is equal to zero or u and v are simultaneously equal to zero (u is the frequency in x-direction, and v is the frequency in y-direction), the operators cannot be sampled at these frequencies. Consequently, the transformation cannot be implemented by fast Fourier transform technique directly. In this article, shift sampling theory is employed for resolving this problem. Model test results show that the technique has good accuracy, and the real case of transformation indicates that the computed results agree better with the measured gradients; it demonstrates not only the effective- ness of method but also the reliability of the measured gradients. 展开更多
关键词 frequency domain horizontal gradient vertical gradient transformATION shift sampiing theory.
原文传递
Speech Encryption in Linear Canonical Transform Domain Based on Chaotic Dynamic Modulation
10
作者 Liyun Xu Tong Zhang Chao Wen 《Journal of Beijing Institute of Technology》 EI CAS 2021年第3期295-304,共10页
In order to transmit the speech information safely in the channel,a new speech encryp-tion algorithm in linear canonical transform(LCT)domain based on dynamic modulation of chaot-ic system is proposed.The algorithm fi... In order to transmit the speech information safely in the channel,a new speech encryp-tion algorithm in linear canonical transform(LCT)domain based on dynamic modulation of chaot-ic system is proposed.The algorithm first uses a chaotic system to obtain the number of sampling points of the grouped encrypted signal.Then three chaotic systems are used to modulate the corres-ponding parameters of the LCT,and each group of transform parameters corresponds to a group of encrypted signals.Thus,each group of signals is transformed by LCT with different parameters.Fi-nally,chaotic encryption is performed on the LCT domain spectrum of each group of signals,to realize the overall encryption of the speech signal.The experimental results show that the proposed algorithm is extremely sensitive to the keys and has a larger key space.Compared with the original signal,the waveform and LCT domain spectrum of obtained encrypted signal are distributed more uniformly and have less correlation,which can realize the safe transmission of speech signals. 展开更多
关键词 communication security linear canonical transform transform domain encryption chaotic system
在线阅读 下载PDF
TRANSFORM DOMAIN SMART ANTENNAS ALGORITHM FOR MAI SUPPRESSION
11
作者 LiKe ShiXinhua ZhangEryang 《Journal of Electronics(China)》 2004年第4期289-295,共7页
Multiple Access Interference(MAI) is the major factor that degrades the performance of a CDMA system. In this paper, a novel transform domain algorithm combined with parameter estimation for MAI suppression is propose... Multiple Access Interference(MAI) is the major factor that degrades the performance of a CDMA system. In this paper, a novel transform domain algorithm combined with parameter estimation for MAI suppression is proposed. Compared with the method that combines an adaptive array antenna with parameter estimation for interference suppression, it converges faster with the same Bit Error Rate(BER) performance. 展开更多
关键词 transform domain Interference cancellation Smart antennas
在线阅读 下载PDF
基于Transformer的高光谱图像域适应分类 被引量:1
12
作者 何文强 李照奎 房卓群 《激光杂志》 北大核心 2025年第2期141-148,共8页
针对跨域高光谱图像分类中的光谱偏移和光谱冗余问题,提出了一种基于Transformer的高光谱图像域适应分类方法。该方法结合逐像素高光谱长波段分块策略和基于邻域相关性的中心像元特征提取策略,有效提取高光谱图像中的局部-长程光谱相关... 针对跨域高光谱图像分类中的光谱偏移和光谱冗余问题,提出了一种基于Transformer的高光谱图像域适应分类方法。该方法结合逐像素高光谱长波段分块策略和基于邻域相关性的中心像元特征提取策略,有效提取高光谱图像中的局部-长程光谱相关性特征和中心像元信息,最后通过双分类器架构实现知识的有效迁移。在Houston和YRD数据集上的实验结果证实了所提方法的有效性。本方法的提出为高光谱图像的域适应分类研究提供了新的视角和技术路径。 展开更多
关键词 高光谱图像 遥感 分类 域适应 transformER
原文传递
基于双域卷积Transformer的滚动轴承剩余寿命预测
13
作者 黄亚欣 李艳婷 《机床与液压》 北大核心 2025年第20期1-7,共7页
深度学习方法在滚动轴承剩余寿命预测中广泛应用,但现有深度学习方法无法充分利用历史信息,且缺乏综合利用局部信息与全局信息的能力。针对上述问题,提出双域卷积Transformer用于预测滚动轴承的剩余寿命。使用滚动轴承从开始运转到预测... 深度学习方法在滚动轴承剩余寿命预测中广泛应用,但现有深度学习方法无法充分利用历史信息,且缺乏综合利用局部信息与全局信息的能力。针对上述问题,提出双域卷积Transformer用于预测滚动轴承的剩余寿命。使用滚动轴承从开始运转到预测时刻的完整振动信号作为输入,为模型学习长期退化过程提供充分的历史数据。使用短时傅里叶变换将完整信号从时域变换到时频域,然后通过双域卷积块在时域和频域两个维度上进行卷积操作,提取信号在时频域上的局部特征。最后,使用Transformer编码器提取信号的全局特征,与双域卷积块提取的局部特征形成互补,以丰富模型学习到的退化信息。使用滚动轴承加速退化数据集进行验证,双域卷积Transformer的平均误差相比注意力加强的LSTM与多尺度CNN方法分别降低了34.8%和16.4%,DDCT在轴承全寿命周期上的平均预测精度有一定的提升。同时,通过消融实验验证了双域卷积Transformer关键模块的有效性。 展开更多
关键词 双域卷积 transformER 滚动轴承 剩余寿命预测 时频特征
在线阅读 下载PDF
LAPPED TRANSFORM DOMAIN INTERFERENCE EXCISION USING MODIFIED MEDIAN FILTER
14
作者 李冲泥 胡光锐 《Journal of Shanghai Jiaotong university(Science)》 EI 1999年第2期32-35,共4页
A novel communication receiver which uses lapped transform(LT) incorporating modified median filter(MMF) algorithm was designed for narrow band interference(NBI) excision.Comparing to traditional Fourier Transform,LT ... A novel communication receiver which uses lapped transform(LT) incorporating modified median filter(MMF) algorithm was designed for narrow band interference(NBI) excision.Comparing to traditional Fourier Transform,LT has longer basis vectors,less spectral leakage,thus better frequency resolution.The LT domain MMF algorithm takes full advantages of the direct sequence spread spectrum signal,as well as the characteristics of LT,performs the transform domain filtering twice.The first filtering locates the position of interference and mitigates most of them.The second filtering is performed in a small neighborhood of the located interference.So LT domain MMF algorithm can completely mitigate the interference without distorting the desired signal.The simulation results demonstrate the improved BER(Bit Error Rate)performance and increased robustness of our receiver. 展开更多
关键词 lapped transform MEDIAN filter INTERFERENCE EXCISION SPREAD SPECTRUM communication transform domain EXCISION
在线阅读 下载PDF
Power System Harmonic Detection Using Frequency-Domain Interpolation Wavelet Transform 被引量:1
15
作者 杜天军 陈光 《Journal of Electronic Science and Technology of China》 2005年第3期245-248,共4页
Aiming at harmonic detection, fast Fourier transform can only detect integer harmonics precisely, short time Fourier transform can detect non-integer harmonics with low resolution, and some former wavelet based method... Aiming at harmonic detection, fast Fourier transform can only detect integer harmonics precisely, short time Fourier transform can detect non-integer harmonics with low resolution, and some former wavelet based methods have no aliasing-reduction scheme which result in low measurement precision and poor robustness. A frequency-domain interpolation algorithm to detect harmonics is proposed by choosing Shannon wavelet. Shannon wavelet is an orthogonal wavelet possessing best ideal frequency domain localization ability, it can restrict wavelet abasing but bring about Gibbs oscillation phenomenon simultaneously. An interpolation algorithm is developed to overcome this problem. Simulation reveals that the proposed method can effectively cancel aliasing, spectral leakage and Gibbs phenomenon, so it provides an effective means for power system harmonic analysis. 展开更多
关键词 wavelet transform frequency-domain interpolation wavelet aliasing harmonic detection
在线阅读 下载PDF
基于Swin-Transformer的多尺度多源域自适应轴承故障诊断 被引量:2
16
作者 周玉国 张志凯 +2 位作者 张金超 于春风 周立俭 《机床与液压》 北大核心 2025年第1期32-42,共11页
针对当前多源域自适应方法无法充分挖掘多源域中不同尺度故障信息的问题,提出一种基于Swin-Transformer(Swin-T)的多尺度多源域自适应轴承故障诊断方法。通过连续小波变换,获得振动信号在不同频带的特征。为更充分地利用多源域中不同尺... 针对当前多源域自适应方法无法充分挖掘多源域中不同尺度故障信息的问题,提出一种基于Swin-Transformer(Swin-T)的多尺度多源域自适应轴承故障诊断方法。通过连续小波变换,获得振动信号在不同频带的特征。为更充分地利用多源域中不同尺度的故障信息,提出基于Swin-T的多尺度特征提取网络。为了减小各域之间的数据分布差异,构建基于最大均值差异的特征对齐网络,并根据不同尺度对分类的贡献赋予权值。此外,构建多尺度特征融合模块,对不同尺度的特征信息进行融合,得到故障特征集。最后,利用Softmax对特征集进行故障分类,并通过最小化多分类器预测差异损失得到最终分类结果。在凯斯西储大学和青岛理工大学轴承数据集上,该方法的故障分类准确度分别达到99.63%和99.40%。 展开更多
关键词 轴承 故障诊断 多源域自适应 Swin-transformer 多尺度特征提取 最大均值差异
在线阅读 下载PDF
基于双域增强Transformer的图像超分辨率重建
17
作者 傅慧滢 杨高明 王瑜 《哈尔滨商业大学学报(自然科学版)》 2025年第2期143-151,共9页
针对现有图像超分辨率方法捕捉图像细节信息不充分导致生成图片质量不佳的问题,提出基于双域增强Transformer(DDET)的图像超分辨率重建方法.该算法从空间域信息学习和频域信息学习两个角度设计模型,通过交替连接空间域增强Transformer模... 针对现有图像超分辨率方法捕捉图像细节信息不充分导致生成图片质量不佳的问题,提出基于双域增强Transformer(DDET)的图像超分辨率重建方法.该算法从空间域信息学习和频域信息学习两个角度设计模型,通过交替连接空间域增强Transformer模块(SETB)和频域增强Transformer模块(FETB),在提取空间域信息的同时有效学习频域信息,增强网络信息提取能力.此外,为了使网络充分关注全局和局部信息,设计特殊的卷积结构与频域信息提取模块融合,进一步提高重建图像质量.相较于基于多尺度残差网络的图像超分辨率(MSRN),当放大倍数为3时,DDET在基准数据集Set5、Set14、Urban100、BSD100上,峰值信噪比(R PSN)指标分别提升0.37、0.22、0.69、0.18 dB;视觉对比上,DDET生成图片纹理更清晰.实验结果表明,DDET可以关注到更多细节信息,生成更高质量的图像,表现出更优越的性能. 展开更多
关键词 图像超分辨率 transformER 频域 傅里叶变换 深度学习 注意力机制
在线阅读 下载PDF
基于双域联合Transformer的压缩感知高光谱重建
18
作者 马超 廉玉生 +3 位作者 张唤唤 刘瑜 王媛媛 曹栩珩 《印刷与数字媒体技术研究》 北大核心 2025年第6期353-361,共9页
高光谱成像技术为同色同谱印刷提供了关键技术支持,尤其在印前图像光谱颜色精准匹配和印刷质量检测方面具有显著优势。双相机压缩感知计算高光谱成像技术具有硬件成本低、可实现快照成像等优点,已经成为高光谱成像领域的研究热点之一。... 高光谱成像技术为同色同谱印刷提供了关键技术支持,尤其在印前图像光谱颜色精准匹配和印刷质量检测方面具有显著优势。双相机压缩感知计算高光谱成像技术具有硬件成本低、可实现快照成像等优点,已经成为高光谱成像领域的研究热点之一。其光谱重建方法是决定目标光谱图像质量和精度的核心关键技术。现有的双相机压缩感知光谱重建算法忽视了HSI的频域特征和局部-全局特征重建,造成了空谱失真和局部细节丢失。为解决上述问题,本研究提出一种基于双域联合Transformer的压缩感知高光谱重建方法(Dual-domain Collaborative Transformer,DDCT),其核心双域联合模块(Dual-domain Collaborative Module,DCM)通过双域交叉注意力机制(Dual-domain Cross-Attention,DCA)实现图像域与频域特征的联合重建;设计局部空频融合模块(Local Spatial-Frequency Fusion Module,LSFM),利用大核卷积和频域融合模块重建局部空间和频域特征,弥补了DCA对于局部特征建模能力的不足。在CAVE与KAIST数据集上的实验表明,与现有的9种算法相比,本研究提出的方法取得了最优的光谱图像重建结果。 展开更多
关键词 压缩感知高光谱成像 光谱重建 transformER 双域联合 频率域特征
在线阅读 下载PDF
变负载条件下电机故障的Transformer-DANN诊断方法研究 被引量:1
19
作者 王嘉铭 蔡浩原 +2 位作者 柳雅倩 邓智瀚 陈骏彬 《控制与决策》 北大核心 2025年第10期3096-3105,共10页
针对深度学习中标准训练集无法全面覆盖实际工况中的故障特征,导致故障识别率急剧下降的问题,提出一种融合一维卷积神经网络(1DCNN)和Transformer层的域对抗神经网络(DANN)迁移学习方法Transformer-DANN.使用1DCNN和Transformer层,改进... 针对深度学习中标准训练集无法全面覆盖实际工况中的故障特征,导致故障识别率急剧下降的问题,提出一种融合一维卷积神经网络(1DCNN)和Transformer层的域对抗神经网络(DANN)迁移学习方法Transformer-DANN.使用1DCNN和Transformer层,改进特征提取器的提取特征的能力,降低计算复杂度;针对不同负载下故障数据特征不同的问题,采用DANN方法对故障数据进行分类处理.对所提出方法进行实验验证,在电机变工况条件下,平均识别率达到98.13%,最大识别率为99.42%.结果表明,所提出方法能有效提高变工况条件下的电机故障识别准确率,可以满足现实应用中设备故障诊断的任务需求. 展开更多
关键词 电机故障诊断 变工况 域对抗神经网络 transformER 特征提取 深度学习
原文传递
基于空频协同的CNN-Transformer多器官分割网络
20
作者 王梦溪 雷涛 +3 位作者 姜由涛 刘乐 刘少庆 王营博 《智能系统学报》 北大核心 2025年第5期1266-1280,共15页
针对目前主流的医学多器官分割网络未能充分利用卷积神经网络(convolutional neural network,CNN)的局部细节提取优势以及Transformer的全局信息捕获潜力,并缺乏空频特征协同建模的问题,提出了一种基于空频协同的CNN-Transformer双分支... 针对目前主流的医学多器官分割网络未能充分利用卷积神经网络(convolutional neural network,CNN)的局部细节提取优势以及Transformer的全局信息捕获潜力,并缺乏空频特征协同建模的问题,提出了一种基于空频协同的CNN-Transformer双分支编解码网络。该网络在局部分支中设计了空频协同注意力,使网络从频域和空间域捕获到更为丰富的局部细节信息;在全局分支设计了多视图频域提取器,该模块通过频谱层和自注意力层联合建模,提高了模型的空频特征协同建模能力和泛化性能。此外,设计了局部与全局特征融合模块,有效整合了CNN分支的局部细节信息和Transformer分支的全局信息,解决了网络无法兼顾局部细节和全局感受野的难题。实验结果表明,该架构克服了医学图像中器官边界模糊导致误分割的问题,有效提升了多器官分割精度,同时计算成本更低,参数量更少。 展开更多
关键词 多器官分割 空频协同 多视图频域 注意力机制 CNN transformER 协同注意力 局部−全局特征融合
在线阅读 下载PDF
上一页 1 2 101 下一页 到第
使用帮助 返回顶部