期刊文献+
共找到9,255篇文章
< 1 2 250 >
每页显示 20 50 100
基于Transformer多分辨率特征融合的图像压缩感知重构
1
作者 熊承义 马帅 +2 位作者 高志荣 李帆 陈文旗 《中南民族大学学报(自然科学版)》 2025年第3期400-406,共7页
利用图像多分辨率特征的交叉融合,对于改善压缩感知图像的重构质量具有较好潜能.研究了一种基于Transformer多分辨率特征融合的图像压缩感知重构方法.输入图像的测量值首先经过初始重构,得到一组分辨率降维的低分辨率初始重构图像;然后... 利用图像多分辨率特征的交叉融合,对于改善压缩感知图像的重构质量具有较好潜能.研究了一种基于Transformer多分辨率特征融合的图像压缩感知重构方法.输入图像的测量值首先经过初始重构,得到一组分辨率降维的低分辨率初始重构图像;然后,采用两个通路并行提取不同分辨率图像的特征并进行交叉融合;最后,将输出的两路特征分别用于原始图像的重构及其降采样重构.采用Transformer网络执行多分辨率图像特征的交叉融合,以更好利用图像的远距离相关性.大量实验比较结果验证了所提出的方法在平衡网络复杂度和改进重构图像质量方面的有效性. 展开更多
关键词 多分辨率特征 压缩感知 交叉融合 transformer方法
在线阅读 下载PDF
基于CNN-Transformer的自由泳划臂动作识别方法
2
作者 舒小东 《通化师范学院学报》 2025年第6期40-47,共8页
为了提升自由泳划臂动作识别性能,利用卷积神经网络(Convolutional Neural Network,CNN)的特征提取能力和Transformer的序列数据处理能力,提出基于CNN-Transformer的自由泳划臂动作识别方法.基于时间能量金字塔算法,利用自由泳划臂动作... 为了提升自由泳划臂动作识别性能,利用卷积神经网络(Convolutional Neural Network,CNN)的特征提取能力和Transformer的序列数据处理能力,提出基于CNN-Transformer的自由泳划臂动作识别方法.基于时间能量金字塔算法,利用自由泳划臂动作图像的深度运动序列能量,计算每层金字塔的深度运动序列能量直方图,提取自由泳划臂动作特征.以GoogleNet模型为基础构建卷积神经网络,对自由泳划臂动作特征图实施多层卷积和池化操作.通过平均池化方法矢量化处理自由泳划臂动作特征,利用卷积核实现划臂动作特征映射.Transformer模块通过多头自注意力机制,捕捉所映射特征的长期依赖关系,明确动作细节,输出自由泳划臂动作识别结果.实验结果表明:该方法能够精准识别入水、抓水、划水等自由泳划臂动作,对于自由泳训练和比赛分析具有重要的应用价值. 展开更多
关键词 CNN transformER 自由泳 划臂动作 识别方法 特征映射
在线阅读 下载PDF
基于近邻感知Transformer的肺孤立性小结节诊断研究
3
作者 侯永泉 王璟琛 +4 位作者 刘瑞 贺燕林 韩佳悦 徐毛冶 柴军 《计算机与数字工程》 2025年第5期1427-1433,共7页
论文探讨了近邻感知Transformer对于诊断肺内孤立性小结节的价值。为了挖掘输入影像不同区域之间的内在关联,论文设计了一种近邻感知Transformer用于肺孤立性小结节的常规诊断。通过回顾性分析2013年1月至2020年5月在内蒙古自治区人民... 论文探讨了近邻感知Transformer对于诊断肺内孤立性小结节的价值。为了挖掘输入影像不同区域之间的内在关联,论文设计了一种近邻感知Transformer用于肺孤立性小结节的常规诊断。通过回顾性分析2013年1月至2020年5月在内蒙古自治区人民医院行胸部CT检查发现肺内孤立性小结节的患者资料97例,根据病理结果分为良性组及恶性组,其中良性40例,恶性57例。利用基于近邻感知Transformer的深度学习模型自动测量结节的体积、平均径、峰度、偏度、最小CT值等参数,比较各参数的诊断效能,绘制ROC曲线确定最佳临界值。实验结果为良性和恶性结节的体积、峰度、最小CT值、平均CT值、25%、50%、75%、90%位数CT值有统计学意义(p<0.05)。体积的最佳界值为1437.7 mm3,AUC为0.62;峰度(Hu)的最佳界值为-1.025,AUC为0.62;最小CT值的最佳界值为-386 Hu,AUC为0.746;平均CT值的最佳界值为-115 Hu,AUC为0.743;25%、50%、75%、90%位数CT值的最佳界值、AUC分别为(-355.5 Hu、0.784,-129.5 Hu、0.756,-3 Hu、0.754,54.5Hu、0.696)。大量实验结果表明近邻感知Transformer方法对肺孤立性小结节良恶性的预判有一定的价值。 展开更多
关键词 transformER 深度学习 体积密度法 孤立性肺小结节
在线阅读 下载PDF
基于分解技术的IZOA-Transformer-BiGRU短期风电功率预测 被引量:2
4
作者 蒲晓云 杨靖 +1 位作者 杨兴 宁媛 《电子测量技术》 北大核心 2025年第2期39-48,共10页
准确的风电功率预测对于保障电网平稳运行和提升风资源利用效率具有重要意义。针对风电功率数据的非平稳性和间歇性等特征,本文提出了一种结合数据分解技术的IZOA-Transformer-BiGRU组合预测模型,以提升短期风电功率预测的精度和可靠性... 准确的风电功率预测对于保障电网平稳运行和提升风资源利用效率具有重要意义。针对风电功率数据的非平稳性和间歇性等特征,本文提出了一种结合数据分解技术的IZOA-Transformer-BiGRU组合预测模型,以提升短期风电功率预测的精度和可靠性。首先,采用能量差值法确定变分模态分解(VMD)的子模态数,将具有较强随机波动性的原始风电功率分解为一系列相对平稳的子序列,从而更加充分地提取时序特征。其次,构建Transformer-BiGRU模型,引入多头注意力机制并行处理多个特征之间的交互关系,并利用BiGRU捕捉时序序列间的前后依赖性,从而提升预测性能。为了进一步优化模型性能,采用融合Singer混沌映射、透镜折射反向学习和单纯形法策略的改进斑马优化算法(IZOA),对Transformer-BiGRU模型的隐藏层神经元数、初始学习率、正则化系数和多头注意力头数四个关键超参数进行优化。最后,通过IZOA-Transformer-BiGRU对分解后的各子序列进行预测,经过叠加重构得到最终的预测结果。实验结果表明,与单一BiGRU模型相比,所提模型的决定系数提升了5.10%,平均绝对误差、均方根误差以及平均绝对百分比误差分别降低了56.17%、54.58%、54.55%,具有较高的预测精度。 展开更多
关键词 风电功率预测 变分模态分解 transformER 双向门控循环单元 能量差值法 斑马优化算法
原文传递
Application of Elzaki Transform Method to Market Volatility Using the Black-Scholes Model
5
作者 Henrietta Ify Ojarikre Ideh Rapheal Ebimene James Mamadu 《Journal of Applied Mathematics and Physics》 2024年第3期819-828,共10页
Black-Scholes Model (B-SM) simulates the dynamics of financial market and contains instruments such as options and puts which are major indices requiring solution. B-SM is known to estimate the correct prices of Europ... Black-Scholes Model (B-SM) simulates the dynamics of financial market and contains instruments such as options and puts which are major indices requiring solution. B-SM is known to estimate the correct prices of European Stock options and establish the theoretical foundation for Option pricing. Therefore, this paper evaluates the Black-Schole model in simulating the European call in a cash flow in the dependent drift and focuses on obtaining analytic and then approximate solution for the model. The work also examines Fokker Planck Equation (FPE) and extracts the link between FPE and B-SM for non equilibrium systems. The B-SM is then solved via the Elzaki transform method (ETM). The computational procedures were obtained using MAPLE 18 with the solution provided in the form of convergent series. 展开更多
关键词 Elzaki transform method European Call Black-Scholes Model Fokker-Planck Equation Market Volatility
在线阅读 下载PDF
A Comprehensive Study of Soliton Pulse Propagation Using Modified Differential Transform Methods
6
作者 Entsar El-Shazly Ahmed Abo-Elenin Ibrahim El-Kalla 《Journal of Applied Mathematics and Physics》 2024年第12期4183-4197,共15页
In this article, a modified version of the Differential Transform Method (DTM) is employed to examine soliton pulse propagation in a weakly non-local parabolic law medium and wave propagation in optical fibers. This s... In this article, a modified version of the Differential Transform Method (DTM) is employed to examine soliton pulse propagation in a weakly non-local parabolic law medium and wave propagation in optical fibers. This semi-analytic method has the advantage of overcoming the obstacle of the hardest nonlinear terms and is used to explain the origin of the bright and dark soliton solutions through the Schrödinger equation in its non-local form and the Radhakrishnan-Kundu-Laksmannan (RKL) equation. Numerical examples demonstrate the effectiveness of this method. 展开更多
关键词 Soliton Pulse Parabolic Law Medium Differential transform method Bright Solution Dark Solution Schrödinger Equation Radhakrishnan-Kundu-Laksmannan
在线阅读 下载PDF
Lite-IJformer:面向长序列Transformer的轻量化方法
7
作者 连家诚 郝一帆 +2 位作者 张曦珊 支天 孙广中 《高技术通讯》 北大核心 2025年第2期167-174,共8页
针对面向长序列的Transformer计算复杂度高的问题,本文提出了一种Trans-former轻量化方法Lite-IJformer。其核心思路分为2步:(1)对自注意力(self-attention)进行线性化,将Transformer的计算复杂度从输入序列的平方降至线性;(2)基于低秩... 针对面向长序列的Transformer计算复杂度高的问题,本文提出了一种Trans-former轻量化方法Lite-IJformer。其核心思路分为2步:(1)对自注意力(self-attention)进行线性化,将Transformer的计算复杂度从输入序列的平方降至线性;(2)基于低秩矩阵分解理论对KV矩阵乘法进行降维,进一步减少计算规模。在长序列竞技基准测试上的实验表明,当输入长度为1000~2000时,线性化可以将self-attention计算量降低13~26倍,将Transformer的推理速度提升4.75~5.72倍而无精度损失;在经过降维后,self-attention的计算量进一步减少了17.0%,模型推理速度提升了1.17倍,精度损失在0.5%以内。 展开更多
关键词 transformER 自注意力 线性化方法 降维
在线阅读 下载PDF
基于序贯三支掩码和注意力融合的Transformer解释方法
8
作者 成晓天 丁卫平 +3 位作者 耿宇 黄嘉爽 鞠恒荣 郭静 《计算机研究与发展》 北大核心 2025年第11期2737-2751,共15页
Transformer逐渐成为计算机视觉任务的首选方案,这推动了其可解释性方法的发展.传统解释方法大多采用Transformer编码器的最终层生成的扰动掩码生成可解释图,而忽略了掩码的不确定信息和上下采样中的信息丢失,从而造成物体区域的定位粗... Transformer逐渐成为计算机视觉任务的首选方案,这推动了其可解释性方法的发展.传统解释方法大多采用Transformer编码器的最终层生成的扰动掩码生成可解释图,而忽略了掩码的不确定信息和上下采样中的信息丢失,从而造成物体区域的定位粗糙且不完整.为克服上述问题,提出基于序贯三支掩码和注意力融合的Transformer解释方法(SAF-Explainer),SAF-Explainer主要包含序贯三支掩码(sequential three-way mask,S3WM)模块和注意力融合(attention fusion,AF)模块.S3WM通过应用严格的阈值条件处理掩码,避免掩码中的不确定信息对解释结果产生损害,以此有效定位到物体位置.随后,AF利用注意力矩阵聚合生成跨层信息交互的关系矩阵,用来优化解释结果中的细节信息,生成边缘清晰且完整的解释结果.为验证所提出SAF-Explainer的有效性,在3个自然图像与1个医学图像数据集上进行比较实验,结果表明SAF-Explainer具有更好的可解释性效果. 展开更多
关键词 可解释性方法 transformER 自注意力机制 序贯三支决策 注意力融合 扰动掩码
在线阅读 下载PDF
Analysis of a Laplace Spectral Method for Time-Fractional Advection-Diffusion Equations Incorporating the Atangana-Baleanu Derivative
9
作者 Kamran Farman Ali Shah +3 位作者 Kallekh Afef J.F.Gómez-Aguilar Salma Aljawi Ioan-Lucian Popa 《Computer Modeling in Engineering & Sciences》 2025年第6期3433-3462,共30页
In this article,we develop the Laplace transform(LT)based Chebyshev spectral collocation method(CSCM)to approximate the time fractional advection-diffusion equation,incorporating the Atangana-Baleanu Caputo(ABC)deriva... In this article,we develop the Laplace transform(LT)based Chebyshev spectral collocation method(CSCM)to approximate the time fractional advection-diffusion equation,incorporating the Atangana-Baleanu Caputo(ABC)derivative.The advection-diffusion equation,which governs the transport of mass,heat,or energy through combined advection and diffusion processes,is central to modeling physical systems with nonlocal behavior.Our numerical scheme employs the LT to transform the time-dependent time-fractional PDEs into a time-independent PDE in LT domain,eliminating the need for classical time-stepping methods that often suffer from stability constraints.For spatial discretization,we employ the CSCM,where the solution is approximated using Lagrange interpolation polynomial based on the Chebyshev collocation nodes,achieving exponential convergence that outperforms the algebraic convergence rates of finite difference and finite element methods.Finally,the solution is reverted to the time domain using contour integration technique.We also establish the existence and uniqueness of the solution for the proposed problem.The performance,efficiency,and accuracy of the proposed method are validated through various fractional advection-diffusion problems.The computed results demonstrate that the proposed method has less computational cost and is highly accurate. 展开更多
关键词 Laplace transform spectral method existence theory fractional derivative with non-singular kernel contour integration methods
在线阅读 下载PDF
Phase-field modeling of effect of Ni on formation and phase transformation of Cu-rich phase in Fe-Cu-Ni alloys
10
作者 Ming-Guang Wei Zhong-Wen Zhang +2 位作者 Min Cui Yuan-Bin Zhang Tong-Guang Zhai 《Chinese Physics B》 2025年第8期735-741,共7页
A phase-field model integrated with the thermodynamic databases was constructed to investigate the impact of Ni content on the precipitation kinetics and phase transformation of the Cu-rich phase in Fe-Cu-Ni alloy at ... A phase-field model integrated with the thermodynamic databases was constructed to investigate the impact of Ni content on the precipitation kinetics and phase transformation of the Cu-rich phase in Fe-Cu-Ni alloy at 773 K.The results demonstrated that the Cu core-Ni shell structures form via the decomposition of Cu-Ni co-clusters,which is consistent with previous experimental results.As the Ni content increases,both the volume fraction and number density of Cu-rich precipitates increase,while their size decreases.With the increase in Ni content,the transformation from a Cu to 9R Cu is accelerated,which is the opposite to the result of increasing Mn content.Magnetic energy can increase the nucleation rate of the Cu-rich phase,but it does not affect the phase transformation driving force required for its crystal structure transformation. 展开更多
关键词 phase-field method Cu-rich phase phase transformation magnetic energy
原文传递
Fast Parallel Magnetic Resonance Imaging Reconstruction Based on Sparsifying Transform Learning and Structured Low-Rank Model
11
作者 DUAN Jizhong XU Yuhan HUANG Huan 《Journal of Shanghai Jiaotong university(Science)》 2025年第3期499-509,共11页
The structured low-rank model for parallel magnetic resonance(MR)imaging can efficiently reconstruct MR images with limited auto-calibration signals.To improve the reconstruction quality of MR images,we integrate the ... The structured low-rank model for parallel magnetic resonance(MR)imaging can efficiently reconstruct MR images with limited auto-calibration signals.To improve the reconstruction quality of MR images,we integrate the joint sparsity and sparsifying transform learning(JTL)into the simultaneous auto-calibrating and k-space estimation(SAKE)structured low-rank model,named JTLSAKE.The alternate direction method of multipliers is exploited to solve the resulting optimization problem,and the optimized gradient method is used to improve the convergence speed.In addition,a graphics processing unit is used to accelerate the proposed algorithm.The experimental results on four in vivo human datasets demonstrate that the reconstruction quality of the proposed algorithm is comparable to that of JTL-based low-rank modeling of local k-space neighborhoods with parallel imaging(JTL-PLORAKS),and the proposed algorithm is 46 times faster than the JTL-PLORAKS,requiring only 4 s to reconstruct a 200×200 pixels MR image with 8 channels. 展开更多
关键词 structured low-rank parallel magnetic resonance imaging sparsifying transform learning alternating direction method of multipliers optimized gradient method
原文传递
Measurement device and method for mass and centroid of large aircraft
12
作者 ZHANG Xiaolin ZHANG Yuyang +2 位作者 YANG Lifeng ZHAO Hongzhi WANG Meibao 《Journal of Measurement Science and Instrumentation》 2025年第3期341-349,共9页
The precise acquisition of the quality characteristic parameters of large aircraft directly affects its performance characteristics.For large aircrafts such as missiles and rockets with internal fillings,traditional m... The precise acquisition of the quality characteristic parameters of large aircraft directly affects its performance characteristics.For large aircrafts such as missiles and rockets with internal fillings,traditional measurement methods involving large-angle tilting or rotation may pose safety risks.In light of the characteristics of large aircraft and in combination with existing measurement methods,we design a mass and centroid measurement method based on four-point support and small-angle tilting,and develop a set of mass and centroid testing system.This method obtains the intersection point of the gravity action line in the product coordinate system through coordinate transformation in two postures,thereby obtaining the three-dimensional centroid of the aircraft.We first elaborate on the principle of this method in detail,then introduce the composition of the equipment,and analyze the structural stress of key components.Finally,experimental verification and uncertainty analysis are carried out.Experimental verification shows that the maximum deviation of the mass measurement accuracy is less than 0.02%,the centroid measurement accuracy in the X direction is±0.15 mm,in the Y direction it is±0.21 mm,and in the Z direction it is±0.19 mm. 展开更多
关键词 large aircraft centroid measurement coordinate transformation multi-point weighing method gravity line
在线阅读 下载PDF
External stress induced pseudo-spinodal transformation pathway with reduced lamellar size in γ-TiAl alloys
13
作者 Jian-Wei Li Tian-Jiao Dong +4 位作者 Chuan-Xin Liang He-Ran Wang Li-Qiang He Tian-Long Zhang Dong Wang 《Rare Metals》 2025年第5期3460-3478,共19页
Reducing the size of the lamellar structures and increasing the number of twin structures are both effective strategies for enhancing the ductility and fracture toughness ofγ-TiAl alloys.Hot isostatic pressing combin... Reducing the size of the lamellar structures and increasing the number of twin structures are both effective strategies for enhancing the ductility and fracture toughness ofγ-TiAl alloys.Hot isostatic pressing combined with heat treatment is an promising method to optimize the microstructure of TiAl alloys and improve their mechanical properties.However,systematic investigations into the microstructural evolution under high temperature pressure/external stress are limited.In this study,by integrating phase field simulations and CALPHAD thermodynamic database,a unique microstructural response to external stress during aging process is revealed.With the increase of external stress,the size of the lamellar structure initially decreases but then increases,while the number of twin structures initially rises but then decreases,showing nonlinear relationships.An increase in external stress shifts the free energy curves,altering the position of c0(the intersection position between free energies ofα_(2)andγ),which leads to a change in the nucleation mechanism from classical nucleation to pseudo-spinodal decomposition and influences the final microstructure ofγprecipitates.Further simulations indicate a linear correlation between optimal external stress and varying Al content.A deeper analysis indicates that the observed variations in the size and twin structures can be attributed to the interplay among the growth rate of existing variants,the competitive nucleation rates of twinned variants and the redistribution of composition under different external stresses.Our findings provide new insights into optimizing microstructures by pressure/external stress in precipitation processes. 展开更多
关键词 Phase field method γ-TiAl alloys Phase transformation Pseudo-spinodal decomposition Autocatalytic effect
原文传递
基于Transformer神经网络的钻井工况识别方法
14
作者 魏刚 肖林 +3 位作者 李爽 李明明 高耸 高小永 《石油钻探技术》 北大核心 2025年第4期33-41,共9页
为解决传统钻井工况识别方法依赖专家经验与阈值判断、忽视时序依赖且类别和精度受限的问题,提出了一种基于Transformer神经网络的钻井工况高效识别方法。该方法利用多头自注意力机制和残差连接提取时序数据的长时序列依赖特征,并在编... 为解决传统钻井工况识别方法依赖专家经验与阈值判断、忽视时序依赖且类别和精度受限的问题,提出了一种基于Transformer神经网络的钻井工况高效识别方法。该方法利用多头自注意力机制和残差连接提取时序数据的长时序列依赖特征,并在编码器输出端接入前馈全连接分类器进行工况分类;训练采用交叉熵损失和Adam优化器,以得到最优模型参数。利用海上钻井历史数据集对比了CNN,LSTM,CNN−LSTM和Transformer等4种模型的性能,结果表明:所有模型损失曲线均能收敛,但CNN模型和CNN−LSTM模型在100次迭代内收敛缓慢且波动较大,而Transformer模型可迅速稳定收敛;加入衍生特征参数后进一步提升了模型的识别精度与泛化能力。研究结果验证了识别方法在长时序多类别特征捕获方面的优势,为钻井作业工况的高效精准识别提供了创新解决方案,对提升钻井监督智能化水平具有重要工程价值。 展开更多
关键词 钻井工况 识别方法 特征提取 时间依赖性 transformer模型
在线阅读 下载PDF
基于特征融合Swin-Tiny Transformer的γ能谱识别方法
15
作者 顾威 孟献才 洪兵 《哈尔滨商业大学学报(自然科学版)》 2025年第2期161-168,共8页
核素识别作为当前核安全研究方向的焦点,其检测的准确性对于安全防护具有重要意义.为了解决放射性核素识别速度慢,准确率不高的问题,提出一种基于特征融合Swin-Tiny Transformer轻量级模型的核素识别方法.通过NaI探测器测量^(133)Ba、^(... 核素识别作为当前核安全研究方向的焦点,其检测的准确性对于安全防护具有重要意义.为了解决放射性核素识别速度慢,准确率不高的问题,提出一种基于特征融合Swin-Tiny Transformer轻量级模型的核素识别方法.通过NaI探测器测量^(133)Ba、^(60)Co、^(152)Eu、^(137)Cs等4种单个放射性核素的能谱数据,制作训练数据集;预处理阶段,采用格拉姆角场法、希尔伯特曲线法与Chirplet变换法将γ能谱信息转化二维图像,并运用特征融合方法更加凸显γ能谱信息特征;设计残差分组卷积模块,通过两个分支分别提取图像的局部和全局特征,并使用残差连接将分支信息进行有效聚合;使用利用NaI探测器采集以上4种放射性核素的混合γ能谱作为测试数据集进行识别验证.实验结果表明,模型的平均识别准确率达到了99.87%,F_(1)分数为99.88%,与其他算法相比,该算法不仅有效提高了放射源的安全防护,避免了辐射威胁,同时在保证识别速度的前提下,进一步提升了识别的准确性. 展开更多
关键词 核素识别 特征融合 残差分组卷积 格拉姆角场法 Swin-Tiny transformer CHIRPLET
在线阅读 下载PDF
Establishment of a New Method for Genetic Transformation of Dunaliella salina 被引量:1
16
作者 柴晓杰 靳非 +2 位作者 丛玉婷 岳金荣 刘艺琼 《Agricultural Science & Technology》 CAS 2017年第8期1374-1377,共4页
The exogenous gene was integrated into Dunaliella salina successfully by using LiAc/PEG mediating method for the first time. According to the results of histochemical staining, transgenic D. salina was blue, showing t... The exogenous gene was integrated into Dunaliella salina successfully by using LiAc/PEG mediating method for the first time. According to the results of histochemical staining, transgenic D. salina was blue, showing that the exogenous GUS gene was successfully expressed in the cells of D. salina. Meanwhile, the effects of growth state of D. salina, plasmid concentration and temperature on its transformation efficiency were studied, and the transformation conditions were optimized. The results show that the optimum conditions for the genetic transformation of D. salina are shown as follows: D. salina was in the early logarithmic phase; plasmid DNA concentration was 600 μg/ml; temperature was 29 ℃, and transformation efficiency was up to 74.8‰ under the best conditions. According to the results of PCR amplification and PCR-Southern hybridization, the target gene had been integrated into genome and was hereditary. 展开更多
关键词 Dunaliela salina LiAc/PEG mediating method GUS gene Genetic transformation
在线阅读 下载PDF
基于残差U-Net和自注意力Transformer编码器的磁场预测方法 被引量:9
17
作者 金亮 尹振豪 +2 位作者 刘璐 宋居恒 刘元凯 《电工技术学报》 EI CSCD 北大核心 2024年第10期2937-2952,共16页
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... 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。 展开更多
关键词 有限元方法 电磁场 深度学习 U-Net transformER
在线阅读 下载PDF
Leaching and Transformation of Nitrogen Fertilizers in Soil After Application of N with Irrigation: A Soil Column Method 被引量:60
18
作者 ZHOU Jian-Bin XI Jin-Gen +1 位作者 CHEN Zhu-Jun LI Sheng-Xiu 《Pedosphere》 SCIE CAS CSCD 2006年第2期245-252,共8页
A soil column method was used to compare the effect of drip fertigation (the application of fertilizer through drip irrigation systems, DFI) on the leaching loss and transformation of urea-N in soil with that of surfa... A soil column method was used to compare the effect of drip fertigation (the application of fertilizer through drip irrigation systems, DFI) on the leaching loss and transformation of urea-N in soil with that of surface fertilization combined with flood irrigation (SFI), and to study the leaching loss and transformation of three kinds of nitrogen fertilizers (nitrate fertilizer, ammonium fertilizer, and urea fertilizer) in two contrasting soils after the fertigation. In comparison to SFI, DFI decreased leaching loss of urea-N from the soil and increased the mineral N (NH4+-N + NO3- -N) in the soil. The N leached from a clay loam soil ranged from 5.7% to 9.6% of the total N added as fertilizer, whereas for a sandy loam soil they ranged between 16.2% and 30.4%. Leaching losses of mineral N were higher when nitrate fertilizer was used compared to urea or ammonium fertilizer. Compared to the control (without urea addition), on the first day when soils were fertigated with urea, there were increases in NH4+-N in the soils. This confirmed the rapid hydrolysis of urea in soil during fertigation. NH4+-N in soils reached a peak about 5 days after fertigation, and due to nitrification it began to decrease at day 10. After applying NH4+-N fertilizer and urea and during the incubation period, the mineral nitrogen in the soil decreased. This may be related to the occurrence of NH4+-N fixation or volatilization in the soil during the fertigation process. 展开更多
关键词 FERTIGATION irrigation method N leaching N transformation nitrogen fertilizer
在线阅读 下载PDF
A modified solution combustion method to superfine Gd_2O_3:Eu^(3+) phosphor:preparation,phase transformation and optical properties 被引量:10
19
作者 叶信宇 高文贵 +2 位作者 夏李斌 聂华平 庄卫东 《Journal of Rare Earths》 SCIE EI CAS CSCD 2010年第3期345-350,共6页
Cubic and monoclinic Gd2O3:Eu3+ phosphors in the range of nano-scale and submicron-scale were prepared by a modified solution combustion method.Coexistence of cubic and monoclinic phases was found in the highest lumin... Cubic and monoclinic Gd2O3:Eu3+ phosphors in the range of nano-scale and submicron-scale were prepared by a modified solution combustion method.Coexistence of cubic and monoclinic phases was found in the highest luminescent sample synthesized at 600 oC.In relation to commercial sample,the relative luminescence intensity was 49.8%.The shape of emission spectrum of the sample thus changed and the charge-transfer-state band of excitation spectrum slightly shift toward higher energies.With increasing the anneal... 展开更多
关键词 Gd2O3:Eu3+ phase transformation solution combustion method optical property rare earths
原文传递
New homotopy analysis transform method for solving the discontinued problems arising in nanotechnology 被引量:4
20
作者 M.M.Khader Sunil Kumar S.Abbasbandy 《Chinese Physics B》 SCIE EI CAS CSCD 2013年第11期135-139,共5页
We present a new reliable analytical study for solving the discontinued problems arising in nanotechnology. Such problems are presented as nonlinear differential-difference equations. The proposed method is based on t... We present a new reliable analytical study for solving the discontinued problems arising in nanotechnology. Such problems are presented as nonlinear differential-difference equations. The proposed method is based on the Laplace trans- form with the homotopy analysis method (HAM). This method is a powerful tool for solving a large amount of problems. This technique provides a series of functions which may converge to the exact solution of the problem. A good agreement between the obtained solution and some well-known results is obtained. 展开更多
关键词 discretized mKdV lattice equation nonlinear differential-difference equations Laplace transform homotopy analysis transform method
原文传递
上一页 1 2 250 下一页 到第
使用帮助 返回顶部