期刊文献+
共找到616篇文章
< 1 2 31 >
每页显示 20 50 100
Construction of Double Right-Border Binary Vector Carrying Non-Host Gene Rxo1 Resistant to Bacterial Leaf Streak of Rice 被引量:1
1
作者 Xu Mei-rong XIA Zhi-hui +3 位作者 ZHAI Wen-xue Xu Jian-long ZHOU Yong-li LI Zhi-kang 《Rice science》 SCIE 2008年第3期243-246,共4页
Rxol cloned from maize is a non-host gene resistant to bacterial leaf streak of rice. pCAMBIA1305-1 with Rxo1 was digested with Sca I and NgoM IV and the double right-border binary vector pMNDRBBin6 was digested with ... Rxol cloned from maize is a non-host gene resistant to bacterial leaf streak of rice. pCAMBIA1305-1 with Rxo1 was digested with Sca I and NgoM IV and the double right-border binary vector pMNDRBBin6 was digested with Hpa I and Xma I. pMNDRBBin6 carrying the gene Rxo1 was acquired by ligation of blunt-end and cohesive end. The results of PCR, restriction enzyme analysis and sequencing indicated that the Rxo1 gene had been cloned into pMNDRBBin6. This double right-border binary vector, named as pMNDRBBin6-Rxol, will play a role in breeding marker-free plants resistant to bacterial leaf streak of rice by genetic transformation. 展开更多
关键词 Rxo1 gene double right-border binary vector RICE bacterial leaf streak selectable marker-free plant resistance gene
在线阅读 下载PDF
Construction of Double Cross-over Expression Vector for Chloroplast Multicistron in Brassica napus L. 被引量:1
2
作者 武玉永 姚庆收 马立新 《Agricultural Science & Technology》 CAS 2013年第3期402-406,共5页
[Objective] This study aimed to construct Brassica napus chloroplast multi- cistron double cross-over expression vector, to lay the foundation for the genetic engi- neering research of Brassica napus chloroplast. [Met... [Objective] This study aimed to construct Brassica napus chloroplast multi- cistron double cross-over expression vector, to lay the foundation for the genetic engi- neering research of Brassica napus chloroplast. [Method] Two primers were designed based on the known Brassica napus chloroplast DNA sequences AF267640 and Z50868 in GenBank. By using PCR method, two Brassica napus L. chloroplast DNA fragments were obtained, which were named RbcL and ACCD. The two Brassica na- pus chloroplast DNA homologous fragments were then cloned into plasmid pMD18-T to obtain recombinant plasmid pHBM715. Tandem expression cassette harboring spectinomycin-resistant gene aadA, mannanase gene man and green fluorescent pro- tein gene gfp was cloned into the plasmid pHBM715, thereby constructing Brassica napus chloroplast multicistron double cross-over expression vector pHBM716, which was transformed into Escherichia coil for expression and identification. [Result] Plate qualitative analysis was conducted for the functional identification of expression cas- sette in the constructed Brassica napus chloroplast multicistron double cross-over ex- pression vector, results showed that the three genes of the same multicistron were all expressed in E. coil [Conclusion] This study successfully constructed Brassica napus chloroplast multicistron double cross-over expression vector, which laid the foundation for the genetic engineering of Brassica napus chloroplast. 展开更多
关键词 Brassica napus L. Chloroplast DNA Multicistron double cross-over Expression vector Functional identification
在线阅读 下载PDF
Generating Marker-Free Transgenic Tobacco Plants by Agrobacteriummediated Transformation with Double T-DNA Binary Vector 被引量:6
3
作者 周红艳 陈松彪 +3 位作者 李旭刚 肖桂芳 魏晓丽 朱祯 《Acta Botanica Sinica》 CSCD 2003年第9期1103-1108,共6页
We have developed a 'double T-DNA' binary vector system for generating selectable marker-free transgenic plants by Agrobacterium-mediated transformation. The 'double T-DNA' binary vector pDLBRBbarm whi... We have developed a 'double T-DNA' binary vector system for generating selectable marker-free transgenic plants by Agrobacterium-mediated transformation. The 'double T-DNA' binary vector pDLBRBbarm which carried two independent T-DNAs, one containing a selectable marker neomycin phosphotransferase (nptII) gene and the other a bargene, was constructed. Transgenic tobacco (Nicotiana tabacum L.) plants were then produced by Agrobacterium-mediated transformation with this vector. Frequency of the primary transformants co-integrated with npt II gene and bar gene was 59.2%. Segregation of two T-DNA regions was found in 3 out of 4 T-1 lines from co-transformed T-0 plants with nptII and bar PPT-resistant and kanamycin-sensitive plants were in approximate 19.5% of the T-1 plants. The result indicated that this 'double T-DNA' vector system could be a workable approach to generate transgenic plants free from selectable marker genes. Co-transformation of nptII gene and bar gene to plants with mixtures of Agrobacterium tumefaciens strains containing single T-DNA vectors was also tested. Frequency of co-transformed plants was 20.0%-47.7% and relatively low as compared with that of 'double T-DNA' vector system. 展开更多
关键词 plant transformation MARKER-FREE double T-DNA vector TOBACCO
在线阅读 下载PDF
A study on improvement of expression of human G-CSF cDNA transferred with retroviral double-copy vector
4
作者 郭葆玉 张淑英 +2 位作者 徐慧敏 孔宪涛 陆德如 《Journal of Medical Colleges of PLA(China)》 CAS 1994年第4期251-255,共5页
In order to overcome the interference of the internal promoter in retroviral vector with a foreign DNA expression, we used the double-copy vector(DC vector) system to improve the effect of transduced gene expression.H... In order to overcome the interference of the internal promoter in retroviral vector with a foreign DNA expression, we used the double-copy vector(DC vector) system to improve the effect of transduced gene expression.Human G-CSF cDNA was inserted into the Bgl Ⅱ site of the polyclonal sites within the U3 region of the 3'long terminal repeat(3'-LTR) in the vector(N2A). After being identified, the gene was transduced into Ψ-2 packaging cell line by using the electroporation method. Consequently, the gene was duplicated in the infected cells, and transferred to the 5'-LRT, and then placed outside the retroviral transcriptional unit. After two weeks the neomycin resistance positive colonies were grown in the G-418 medium. The supernatant virus was titred and then transferred to NIH3T3 cells. The G-CSF value of positive clone N2AG7-4 was up to 53. 3 U/106 cells. Southern blot and Northern blot analysis showed that the chimeric gene faithfully duplicated in the cells infected with the corresponding virus and generated two copies with one in each LTR. 展开更多
关键词 double-copy vector G-CSF CDNA gene transfer
在线阅读 下载PDF
Detecting Shifted Double JPEG Compression Tampering Utilizing Both Intra-Block and Inter-Block Correlations 被引量:1
5
作者 张玉金 李生红 王士林 《Journal of Shanghai Jiaotong university(Science)》 EI 2013年第1期7-16,共10页
Copy-paste forgery is a very common type of forgery in JPEG images.The tampered patch has always suffered from JPEG compression twice with inconsistent block segmentation.This phenomenon in JPEG image forgeries is cal... Copy-paste forgery is a very common type of forgery in JPEG images.The tampered patch has always suffered from JPEG compression twice with inconsistent block segmentation.This phenomenon in JPEG image forgeries is called the shifted double JPEG(SDJPEG) compression.Detection of SDJPEG compressed image patches can make crucial contribution to detect and locate the tampered region.However,the existing SDJPEG compression tampering detection methods cannot achieve satisfactory results especially when the tampered region is small.In this paper,an effective SDJPEG compression tampering detection method utilizing both intra-block and inter-block correlations is proposed.Statistical artifacts are left by the SDJPEG compression among the magnitudes of JPEG quantized discrete cosine transform(DCT) coefficients.Firstly,difference 2D arrays,which describe the differences between the magnitudes of neighboring JPEG quantized DCT coefficients on the intrablock and inter-block,are used to enhance the SDJPEG compression artifacts.Then,the thresholding technique is used to deal with these difference 2D arrays for reducing computational cost.After that,co-occurrence matrix is used to model these difference 2D arrays so as to take advantage of second-order statistics.All elements of these co-occurrence matrices are served as features for SDJPEG compression tampering detection.Finally,support vector machine(SVM) classifier is employed to distinguish the SDJPEG compressed image patches from the single JPEG compressed image patches using the developed feature set.Experimental results demonstrate the efficiency of the proposed method. 展开更多
关键词 passive image forensics copy-paste forgery shifted double JPEG (SDJPEG) compression co-occurrence matrix support vector machine (SVM)
原文传递
Estimation of Winds at Different Isobaric Levels Based on the Observed Winds at 850 hPa Level Using Double Fourier Series
6
作者 S. N. Bavadekar (1) R. M. Khaladkar (1) 《Advances in Atmospheric Sciences》 SCIE CAS CSCD 1994年第3期327-334,共8页
A technique based on the double Fourier series is developed to estimate the winds at different isobaric levels forthe limited area domain, 35°E to 140°E and 30°S to 40°N, using the observed winds a... A technique based on the double Fourier series is developed to estimate the winds at different isobaric levels forthe limited area domain, 35°E to 140°E and 30°S to 40°N, using the observed winds at 850 hPa lcvcl for the month ofJune. For this purpose the wind field at a level under consideration is taken in the ratio form with that of 850 hPa level and the coefficients of the double Fouricr series are computed. These coefficients are subsequently used to computethe winds which are compared with the actual winds. The results of the double Fourier series technique are comparedwith those of the polynomial surface fitting method developed by Bavadekar and Khaladkar (1 992). The technique isalso applied for the daily wind data of 11. June, 1979 and the validation of the technique is tested for a few radiosondestations of india. The computed winds for these radiosonde stations arc quite close to observed winds. 展开更多
关键词 double Fourier series Objective analysis Cloud motion vectors Numerical weather prediction Polynomial surface fitting
在线阅读 下载PDF
The Double Excitation: A Solution to Improve Energetic Performances of High Power Synchronous Machines
7
作者 Aymen Ammar Frederic Gillon Pascal Brochet 《Journal of Energy and Power Engineering》 2012年第5期719-724,共6页
DFSM (doubly fed synchronous machine) presents several advantages such as efficiency improvement, weight reduction and increase of the utilization factor (kW/kg). In this paper the authors focus on impact of the D... DFSM (doubly fed synchronous machine) presents several advantages such as efficiency improvement, weight reduction and increase of the utilization factor (kW/kg). In this paper the authors focus on impact of the DFSM on the efficiency and machine weight in comparison to conventional synchronous generator with wound rotor. Different topologies of DFSM are briefly described and the different methods and models for performances prediction are presented. 展开更多
关键词 double excitation LOSSES no load characteristic permanent magnet reluctances network vector diagram.
在线阅读 下载PDF
Position Adaptive Measurement Model for Double-four Quadrant Photoelectric Detector
8
作者 党丽萍 刘君华 +1 位作者 唐树刚 刘洪飞 《Defence Technology(防务技术)》 SCIE EI CAS 2006年第2期106-111,共6页
A modeling method of the support vector machine combined with matrix optics is considered; a complete new measurement model for double-four quadrant photoelectric detector is built. According to the analysis of the re... A modeling method of the support vector machine combined with matrix optics is considered; a complete new measurement model for double-four quadrant photoelectric detector is built. According to the analysis of the received light spot size and its motion with the changes of the defocusing amount of detector photosensitive surface and the detector position attitude in the optical path, a mathematic expression of photoelectrical conversion is given, which can be applicable to random setting position of the detector at any time. Based on least square support vector machine (LS SVM), the mapping relationship among the output signal linear characteristic parameters (zero neighborhood gradient and intercept), the defocusing amount of the detector and the installation position attitude angle is established. Thus, the multiple dimensional high accuracy measuring and adjusting control system can be left out, and adaptive measurement of the detector parameters can be realized. Compared with existed measurement model and method, the presented model has the advantages of more clear physical meaning, closer to work mechanism of detector, acquiring more complete sample data and wiping out the dead spots or bad spots in measurement. And the accuracy of displacement measurement is increased to 3?μm. At the same time, this measurement mode provides a technical shortcut for three-dimensional small angle measurement. 展开更多
关键词 光电探测器 位置适应管理 管理模型 无线电导引 线性参数
在线阅读 下载PDF
Genetic Engineering of Tobacco with Double Resistance to Both Virus and Insect
9
作者 梁晓友 米景九 +1 位作者 潘乃隧 陈章良 《Developmental and Reproductive Biology》 1994年第1期29-36,T001,共9页
Expression vector pE14 with double resistance to virus and insect was constructed by inserting CW-cp gene and Bt-toxin gene one by into T-DNA of the same binary vector pE3. Tobacco was then transformed with Agrbacter... Expression vector pE14 with double resistance to virus and insect was constructed by inserting CW-cp gene and Bt-toxin gene one by into T-DNA of the same binary vector pE3. Tobacco was then transformed with Agrbacterium tumefaciens (At)GV311-SE carrying PE14. Nopaline assay and PCR amplification confirmed that both CW-cp gene and Bt-toxin gene had been introduced into tobacco genome by T-DNA of PE3. Test attack with virus and demonstrated, in some of the transgenic plants, double resistance to both infection by CMV and damage by Manduca sexta. 展开更多
关键词 Coiistruction of vector Transformation of tobacco double resistance to virus and insect
在线阅读 下载PDF
不同来流下双楔定常/非定常流动的模拟分析与流态预测
10
作者 张鑫泽 李沁 +1 位作者 翁谊辉 尤延铖 《北京航空航天大学学报》 北大核心 2026年第1期371-384,共14页
针对高超声速双楔绕流中层流激波相互作用和定常/非定常流动现象,选取不同工况(Ma∞和Re),采用所发展的三阶精度本质无波动格式开展数值模拟与流动特性研究,总结由参数变化引起的定常/非定常流场激波结构和分离区的变化规律。为了对不... 针对高超声速双楔绕流中层流激波相互作用和定常/非定常流动现象,选取不同工况(Ma∞和Re),采用所发展的三阶精度本质无波动格式开展数值模拟与流动特性研究,总结由参数变化引起的定常/非定常流场激波结构和分离区的变化规律。为了对不同Ma∞、Re下双楔流动出现的定常/非定常流态进行预测,基于最小二乘支持向量机(LS-SVM)模型,采用不同核函数(线性、多项式和径向基函数)构建双楔绕流定常/非定常的流态预测模型,并在验证集上对3种预测模型进行验证与评估。结果表明:径向基核预测模型能够正确预测16组工况下的双楔流态,预测结果的准确率和真阳性率为1、假阳性率为0,预测能力最强,具有较优的泛化能力,多项式核预测模型预测能力次之,线性核预测模型预测能力最差。 展开更多
关键词 高超声速 双楔 定常/非定常流动 流态预测 最小二乘支持向量机
原文传递
基于速度环双积分控制器的永磁同步电机转子初始位置估计
11
作者 卢洁莹 闻成 +1 位作者 李锦辉 苏为洲 《控制理论与应用》 北大核心 2025年第9期1733-1745,共13页
永磁同步电机转子初始位置估计在实现电机矢量控制、保证电机运行精度方面起着重要作用.针对带增量式编码器的永磁同步电机,文章提出一种基于速度环双积分控制器的转子初始位置估计方法.通过对电机速度环/电流环进行机理分析和辨识,得... 永磁同步电机转子初始位置估计在实现电机矢量控制、保证电机运行精度方面起着重要作用.针对带增量式编码器的永磁同步电机,文章提出一种基于速度环双积分控制器的转子初始位置估计方法.通过对电机速度环/电流环进行机理分析和辨识,得到不同初始电角度估计误差下速度环对象的传递函数.根据这类传递函数的不确定性特点,提出速度环鲁棒控制器设计方案.然后,针对初始位置估计的特殊要求,设计了特定的电机d轴梯形波电流指令信号.根据矢量控制技术中的坐标变换,将d轴指令电流信号转化为q轴电流环的外部输入信号,并将初始位置估计问题转化为速度环的指令跟踪问题.通过设计速度环双积分控制器,实现对电机转子初始位置的精确估计,同时达到低位移低冲击的要求.实验结果表明该方法适用于从电机转子任意不同初始位置进行估计,电角度估计误差小于1:5◦,电机在估计过程中所产生的电角度位移不超过2◦,即具有估计精度高以及估计过程电机产生的位移和冲击小等特点. 展开更多
关键词 初始位置估计 永磁同步电机 双积分控制 矢量控制
在线阅读 下载PDF
帧内编码相同编码参数的VVC双压缩检测算法
12
作者 李强 彭依浪 宋衍昭 《重庆邮电大学学报(自然科学版)》 北大核心 2025年第3期346-356,共11页
为了解决多功能视频编码(versatile video coding,VVC)标准下具有相同编码参数的视频双压缩检测方法准确率不高的问题,提出了一种基于编码单元(coding unit,CU)尺寸、划分模式和预测模式的检测方法。对待检测的视频进行多次编解码,分析... 为了解决多功能视频编码(versatile video coding,VVC)标准下具有相同编码参数的视频双压缩检测方法准确率不高的问题,提出了一种基于编码单元(coding unit,CU)尺寸、划分模式和预测模式的检测方法。对待检测的视频进行多次编解码,分析并确定VVC流中与压缩编码次数密切相关的基础码流特征;以CU尺寸、划分模式和预测模式构建高级码流特征输入支持向量机完成视频的双压缩检测。实验结果表明,与对比文献的方法相比,所提方法的视频双压缩检测准确率有较大提升,平均准确率达到了95.82%。 展开更多
关键词 多功能视频编码 双压缩检测 高级码流特征 支持向量机
在线阅读 下载PDF
一种船舶光学经纬仪的测量误差校准系统的设计与开发
13
作者 徐静 陈岩 《舰船科学技术》 北大核心 2025年第3期148-151,共4页
船舶光学经纬仪能够有效提升船舶导航的精确性,在复杂海况下光学经纬仪的测量存在依赖人工经验且精度低的问题。本文提出一种基于GPS载波相位差分技术结合双矢量定姿技术的光学经纬仪测量误差校准系统,通过实时测量获取船舶姿态角,并将... 船舶光学经纬仪能够有效提升船舶导航的精确性,在复杂海况下光学经纬仪的测量存在依赖人工经验且精度低的问题。本文提出一种基于GPS载波相位差分技术结合双矢量定姿技术的光学经纬仪测量误差校准系统,通过实时测量获取船舶姿态角,并将其补偿到经纬仪测量结果中。设计系统整体结构,并对校准前后的测量误差进行对比。结果表明,校准后水平角和垂直角的误差均在10″以内,满足光学经纬仪测量误差校准的使用要求。 展开更多
关键词 光学经纬仪 误差校准 实时性 双矢量定姿技术 GPS载波相位差分技术
在线阅读 下载PDF
SVPWM技术在PMSM矢量控制系统仿真实验教学中的应用
14
作者 尚丽 黄博 《苏州市职业大学学报》 2025年第2期37-43,69,共8页
分析永磁同步电机的数学模型和矢量控制系统组成,借助MATLAB/Simulink构建基于SVPWM技术的永磁同步电机矢量控制系统仿真模型,并通过设置合理的仿真参数,实现永磁同步电机空载和带载的仿真分析。仿真结果证明了该模型的有效性以及控制... 分析永磁同步电机的数学模型和矢量控制系统组成,借助MATLAB/Simulink构建基于SVPWM技术的永磁同步电机矢量控制系统仿真模型,并通过设置合理的仿真参数,实现永磁同步电机空载和带载的仿真分析。仿真结果证明了该模型的有效性以及控制算法的正确性,为后续的现场教学和调试提供了理论基础和指导思路。 展开更多
关键词 永磁同步电机 空间矢量脉宽调制 双闭环调速 MATLAB/SIMULINK仿真
在线阅读 下载PDF
基于改进梅森旋转算法的永磁同步电机双随机SVPWM高频电磁振动抑制 被引量:2
15
作者 陈浈斐 万向民 +2 位作者 陈书桐 王磊 王枫 《电工技术学报》 北大核心 2025年第8期2504-2517,共14页
在以变频器驱动的永磁同步电机控制系统中,传统的空间矢量脉宽调制(SVPWM)技术会在逆变器开关频率及其倍频处产生高频电流谐波,进而导致永磁同步电机在运行过程中产生较大的高频电磁振动。因此,提出一种随机开关频率与随机零矢量结合的... 在以变频器驱动的永磁同步电机控制系统中,传统的空间矢量脉宽调制(SVPWM)技术会在逆变器开关频率及其倍频处产生高频电流谐波,进而导致永磁同步电机在运行过程中产生较大的高频电磁振动。因此,提出一种随机开关频率与随机零矢量结合的双随机SVPWM控制方法,应用于永磁同步电机高频电流谐波频谱扩展中,该方法可分散集中在开关频率及其倍频处的高频谐波,有效地削弱高频电磁振动。同时,采用改进梅森旋转(MT)算法生成随机数,增强了随机序列的随机性和空间遍历性,保证了双随机SVPWM控制方法的扩频效果。实验结果表明,基于改进MT算法的双随机SVPWM控制方式达到了有效抑制电机高频段电磁振动的目的。 展开更多
关键词 永磁同步电机 电流谐波 电磁振动 双随机空间矢量脉宽调制(SVPWM)
在线阅读 下载PDF
改进的三相并网逆变器双矢量无模型预测控制策略 被引量:1
16
作者 芮涛 徐婉伦 +2 位作者 胡存刚 冯壮壮 陆格野 《电工技术学报》 北大核心 2025年第18期6029-6039,共11页
为解决传统双矢量模型预测控制(DV-MPC)中矢量作用时间计算和预测电流计算对并网逆变器系统参数依赖性强的问题,该文提出一种改进的三相并网逆变器双矢量无模型预测控制(DV-MFPC)策略。首先,分析参数失配对DV-MPC中矢量作用时间以及预... 为解决传统双矢量模型预测控制(DV-MPC)中矢量作用时间计算和预测电流计算对并网逆变器系统参数依赖性强的问题,该文提出一种改进的三相并网逆变器双矢量无模型预测控制(DV-MFPC)策略。首先,分析参数失配对DV-MPC中矢量作用时间以及预测电流计算的影响;其次,提出DV-MFPC策略,所提策略建立并网逆变器数据模型,通过结合最小二乘法,并设计自适应遗忘因子,对不同电压矢量对应的数据模型进行更新;再次,结合双矢量合成方式以及数据模型,实现双矢量无模型电流预测计算,所提方法可以实时辨识并更新三相并网逆变器数据模型,消除了参数对DV-MPC的影响,提高了逆变器数据模型的辨识精度及噪声抑制能力,改善了并网电流性能;最后,通过仿真和实验验证了所提DV-MFPC策略的有效性。 展开更多
关键词 三相并网逆变器 无模型预测控制 数据模型 最小二乘法 双矢量
在线阅读 下载PDF
基于复值卷积降噪自编码器去噪的矢量水听器DOA估计方法 被引量:1
17
作者 任晶 谭秀辉 +3 位作者 白艳萍 王宏妍 续婷 程蓉 《测试技术学报》 2025年第4期475-482,490,共9页
针对现有的基于实值卷积神经网络估计矢量水听器波达方向(Direction of Arrival, DOA)对阵列接收到信号的相位特征提取不充分的缺点,提出了一种基于复值卷积降噪自编码器(Complex-Valued Convolutional Denoising Autoencoder, CV-CDAE... 针对现有的基于实值卷积神经网络估计矢量水听器波达方向(Direction of Arrival, DOA)对阵列接收到信号的相位特征提取不充分的缺点,提出了一种基于复值卷积降噪自编码器(Complex-Valued Convolutional Denoising Autoencoder, CV-CDAE)和复值卷积神经网络(Complex-Valued Convolutional Neural Network, CVCNN)联合的矢量水听器DOA估计方法CV-CDAE-CNN。首先,将矢量水听器接收信号的复值协方差矩阵输入CV-CDAE模块去除噪声,之后,将去噪后的样本输入CV-CNN进行分类。其中,CV-CNN在下采样前使用双尺度膨胀卷积增大特征图的感受野,缓解下采样带来的信息损失。通过CV-CDAE去噪以及CV-CNN独特的处理复值方式实现角度分类,进而得到DOA估计值。仿真结果表明,所提出方法与现有的CV-CNN相比,在低信噪比或有限快拍数下泛化能力更强,DOA估计准确率更高,且具有更高的估计精度。 展开更多
关键词 波达方向估计 复值卷积降噪自编码器 复值卷积神经网络 双尺度膨胀卷积 矢量水听器
在线阅读 下载PDF
面向SW26010-Pro众核处理器的新型矩阵存储格式及稀疏矩阵向量乘(SpMV)算法研究
18
作者 王萃 刘芳芳 +2 位作者 马文静 赵玉文 胡力娟 《计算机学报》 北大核心 2025年第6期1290-1304,共15页
稀疏矩阵向量乘(Sparse Matrix-Vector Multiplication,SpMV)是高性能计算、人工智能大模型领域中的关键操作,其性能通常对应用程序整体性能的提升具有重要影响。高效的稀疏矩阵存储格式是影响SpMV性能的重要因素,然而,现有的稀疏矩阵... 稀疏矩阵向量乘(Sparse Matrix-Vector Multiplication,SpMV)是高性能计算、人工智能大模型领域中的关键操作,其性能通常对应用程序整体性能的提升具有重要影响。高效的稀疏矩阵存储格式是影响SpMV性能的重要因素,然而,现有的稀疏矩阵存储格式主要通过压缩零元素以减少访存,未充分利用非零元素的数值规律,因此仍有进一步压缩和优化的空间。本文通过对压缩稀疏行(Compressed Sparse Row,CSR)存储格式中非零元数组内的重复元素进行进一步的压缩,提出了一种新型的稀疏矩阵存储格式(Further Compressed Sparse Row,FCSR),并设计了从CSR到FCSR格式转换的异构并行算法,以尽量减少格式转换带来的开销。同时,本文面向SW26010-Pro众核处理器,设计了基于FCSR存储格式的SpMV异构并行算法,对SpMV进行了细粒度的任务划分和并行优化设计,探究了五种向量x的间接访存方式,并通过双缓冲技术对算法进行了优化。最后,本文选用SuiteSparse矩阵集中的稀疏矩阵进行了测试,实验结果表明,本文提出的基于FCSR存储格式的异构众核SpMV算法相较于主核版SpMV算法具有明显的性能提升,最高加速比达到43.11,平均加速比为7.56,测试矩阵最高带宽利用率达到了91.13%,平均带宽利用率为26.27%。另外,本文对基于FCSR存储格式和CSR存储格式的SpMV算法性能进行了比较,在两者均得到充分优化的前提下,基于FCSR存储格式的SpMV算法相较于基于CSR存储格式的SpMV算法性能的平均加速比达到1.19。 展开更多
关键词 稀疏矩阵向量乘 SW26010-Pro众核处理器 新型矩阵存储格式 并行优化 双缓冲技术
在线阅读 下载PDF
标准矢量分析法比较平板襻式与双C襻式复曲面人工晶状体对散光矫正的效果
19
作者 夏锦涛 刘佳 +2 位作者 郝咪 马挺 程丽娜 《国际眼科杂志》 2025年第4期632-637,共6页
目的:采用标准矢量分析法比较平板襻式复曲面人工晶状体(AT TORBI 709M)和双C襻式复曲面人工晶状体(Tecnis ZMT)对角膜散光患者术后3 mo的散光矫正效果。方法:回顾性病例对照研究。收录2021-06/12在我院日间手术中心接受超声乳化白内障... 目的:采用标准矢量分析法比较平板襻式复曲面人工晶状体(AT TORBI 709M)和双C襻式复曲面人工晶状体(Tecnis ZMT)对角膜散光患者术后3 mo的散光矫正效果。方法:回顾性病例对照研究。收录2021-06/12在我院日间手术中心接受超声乳化白内障吸除联合散光矫正型人工晶状体植入术治疗的患者69例69眼,把患者按植入不同晶状体分为两组:一组植入AT TORBI 709M(AT TORBI 709M组,38例38眼),一组植入Tecnis ZMT(Tecnis ZMT组,31例31眼)。记录眼轴长度、术前散光和轴位及人工晶状体的度数;术前,术后1 wk,1、3 mo的裸眼远视力(UCDVA)、最佳矫正远视力(BCDVA)、屈光度、残余散光和轴位。采用标准散光矢量分析法评估术后指标,包括等效球镜度(SE)、目标散光矢量(TIA)、散光矢量(SIA)、误差的幅度(ME)、误差角度的绝对值(|AE|)、差异矢量的绝对值(|DV|)、矫正指数(CI)及成功指数(IOS)等指标。结果:两组手术后UCDVA、BCDVA显著提高(均P<0.001),与术前比较有差异(均P<0.001),两组间UCDVA、BCDVA总体比较无差异(P=0.275、0.124)。矢量分析显示,两组都有较好的散光矫正能力,|DV|和IOS均接近0(P=0.329、0.288);AT TORBI 709M组的CI更接近1,更接近预期的矫正散光,Tecnis ZMT组CI>1,散光过矫,两组比较无差异(P=0.193)。术后3 mo,AT TORBI 709M组、Tecnis ZMT组残余散光分别为-0.11±0.91、-0.46±0.76 D,两组无差异(t=1.732,P=0.088)。结论:平板襻式和双C襻式复曲面人工晶状体都可显著提高规则性角膜散光患者的术后视力,具有良好的旋转稳定性,对顺规、逆规散光矫正能力相当。 展开更多
关键词 白内障 标准矢量分析法 角膜散光 平板襻式 双C襻式 旋转稳定性
暂未订购
半周期计算单采样双更新的永磁同步电机低载波比下无位置传感器控制 被引量:1
20
作者 吴春 施淑娟 +1 位作者 朱春桥 郑露华 《兵工学报》 北大核心 2025年第3期289-301,共13页
针对永磁同步电机下桥臂电流采样拓扑的无位置传感器控制系统,在低载波运行时受数字延时影响易导致位置估计精度低、电流纹波增大、稳定性差等问题,提出一种半周期计算、单采样双更新的空间矢量脉宽调制方法,有效减少系统数字延时。该... 针对永磁同步电机下桥臂电流采样拓扑的无位置传感器控制系统,在低载波运行时受数字延时影响易导致位置估计精度低、电流纹波增大、稳定性差等问题,提出一种半周期计算、单采样双更新的空间矢量脉宽调制方法,有效减少系统数字延时。该方法在每个脉冲宽度调制(Pulse-Width Modulation,PWM)的载波过零处采样一次电流,并在采样结束后的半周期内完成目标电压矢量计算。考虑数字延时中电机转过的角度,将目标电压矢量在下一个PWM周期的前、后半周期输出,并分别补偿0.75 T s和1.25 T s等效延时(T s为PWM周期时间),以减小目标电压矢量与实际输出电压矢量之间的误差。将新方法应用于采用模型参考自适应速度观测器的永磁同步电机无位置传感器控制系统中,实验结果表明该方法位置估计精度高、电流纹波小,且可以稳定运行于额定负载、载波比为4.0的工况。 展开更多
关键词 永磁同步电机 无位置传感器控制 低载波比 空间矢量脉宽调制 单采样双更新
在线阅读 下载PDF
上一页 1 2 31 下一页 到第
使用帮助 返回顶部