期刊文献+
共找到6篇文章
< 1 >
每页显示 20 50 100
基于非凸log模型的脑电时-频-空特征选择方法
1
作者 王棋辉 莫云 +2 位作者 梁国富 许川佩 张本鑫 《桂林电子科技大学学报》 2024年第2期153-161,共9页
针对运动想象脑电时-频-空特征选择问题,提出了基于非凸log模型的稀疏特征选择方法(LOG方法)。首先,对原始脑电信号进行时-频分解,得到多个时-频段;其次,针对每个时-频段使用共空域模式(CSP)提取特征,得到时-频-空特征集合;最后,通过提... 针对运动想象脑电时-频-空特征选择问题,提出了基于非凸log模型的稀疏特征选择方法(LOG方法)。首先,对原始脑电信号进行时-频分解,得到多个时-频段;其次,针对每个时-频段使用共空域模式(CSP)提取特征,得到时-频-空特征集合;最后,通过提出的基于log函数的非凸稀疏优化模型进行特征选择和分类,该模型可有效缓解L1范数正则化的有偏估计。为验证本方法的有效性,用3个公开的运动想象脑电数据集进行实验,相比现有的凸稀疏优化模型,非凸log模型取得了82.5%的平均分类准确率。实验结果表明,LOG方法不仅分类准确率高,且模型具有较好的稳定性和鲁棒性。 展开更多
关键词 脑电解码 运动想象 --特征 特征选择 非凸模型
下载PDF
深度稀疏自编码网络融合多LBP特征用于单样本人脸识别
2
作者 赵淑欢 万品哲 郭昌隆 《物联网技术》 2019年第11期13-17,共5页
单样本人脸识别的关键在于充分挖掘单样本判别性信息,采用深度稀疏自编码网络与空频域多LBP特征融合进行特征提取。首先利用部分样本训练深度稀疏自编码网络,利用训练好的网络分别提取训练及测试集的特征;其次,利用二维离散小波变换将... 单样本人脸识别的关键在于充分挖掘单样本判别性信息,采用深度稀疏自编码网络与空频域多LBP特征融合进行特征提取。首先利用部分样本训练深度稀疏自编码网络,利用训练好的网络分别提取训练及测试集的特征;其次,利用二维离散小波变换将时域样本变换到频域,实现样本扩展,增加单样本信息并分别提取各域上的多LBP特征;最后利用协同表示对深度自编码网络及多LBP特征进行分类识别,融合识别结果获取最终分类结果。在AR及PIE数据库上的实验结果表明,该融合算法能提高样本判别性信息的提取,提高单样本人脸识别性能。 展开更多
关键词 稀疏自编码 单样本人脸识别 空-频特征 特征融合 二维离散小波变换 数据库
下载PDF
基于多域特征的滚动轴承故障检测和状态识别方法 被引量:2
3
作者 李大江 《机械设计与制造工程》 2021年第3期55-58,共4页
为提高滚动轴承故障检测与识别效果,提出一种基于局部均值分解(LMD)和共空间模式(CSP)的时-频-空多域特征提取方法。首先采用LMD将滚动轴承信号分解为多个乘积分量(PF)并提取时-频熵特征,基于支持向量数据描述(SVDD)分类器实现正常和故... 为提高滚动轴承故障检测与识别效果,提出一种基于局部均值分解(LMD)和共空间模式(CSP)的时-频-空多域特征提取方法。首先采用LMD将滚动轴承信号分解为多个乘积分量(PF)并提取时-频熵特征,基于支持向量数据描述(SVDD)分类器实现正常和故障轴承的分类;然后利用CSP对故障轴承信号进行分解并提取空域熵特征;最后利用K-均值聚类算法进行聚类,实现对外圈故障、内圈故障和滚柱故障的区分。实验结果表明,所提方法可以获得优于80%的正确分类性能,明显优于传统单一维度特征。 展开更多
关键词 滚动轴承 故障检测 --多域特征 局部均值分解 间模式
下载PDF
SPATIO-TEMPORAL VARIATION CHARACTERISTICS OF EXTREMELY HEAVY PRECIPITATION FREQUENCY OVER SOUTH CHINA IN THE LAST 50 YEARS 被引量:2
4
作者 陆虹 陈思蓉 +2 位作者 郭媛 何慧 徐圣璇 《Journal of Tropical Meteorology》 SCIE 2014年第3期279-288,共10页
This paper comprehensively studies the spatio-temporal characteristics of the frequency of extremely heavy precipitation events over South China by using the daily precipitation data of 110 stations during 1961 to 200... This paper comprehensively studies the spatio-temporal characteristics of the frequency of extremely heavy precipitation events over South China by using the daily precipitation data of 110 stations during 1961 to 2008 and the extremely heavy precipitation thresholds determined for different stations by REOF, trend coefficients, linear trend, Mann-Kendall test and variance analysis. The results are shown as follows. The frequency distribution of extremely heavy precipitation is high in the middle of South China and low in the Guangdong coast and western Guangxi. There are three spatial distribution types of extremely heavy precipitation in South China. The consistent anomaly distribution is the main type. Distribution reversed between the east and the west and between the south and the north is also an important type. Extremely heavy precipitation events in South China mainly occurred in the summer-half of the year. Their frequency during this time accounts for 83.7% of the total frequency. In the 1960 s and 1980 s, extremely heavy precipitation events were less frequent while having an increasing trend from the late 1980 s. Their climatological tendency rates decrease in the central and rise in the other areas of South China, and on average the mean series also shows an upward but insignificant trend at all of the stations. South China's frequency of extremely heavy precipitation events can be divided into six major areas and each of them shows a different inter-annual trend and three of the representative stations experience abrupt changes by showing remarkable increases in terms of Mann-Kendall tests. 展开更多
关键词 South China frequency of extremely heavy precipitation events spatio-temporal characteristics abrupt change
下载PDF
一种基于EEG信号的抑郁症早期筛查方法
5
作者 任书瑶 宋江玲 张瑞 《计算机科学》 CSCD 北大核心 2023年第S02期999-1004,共6页
抑郁症作为一类常见的、可治愈型的精神类疾病,若能在早期阶段对其进行有效筛查(即早期筛查)并及时采取相应的治疗手段,则可有效控制病情的进一步加重,甚至彻底治愈。传统的抑郁症诊断方法主要是医生通过患者的临床表现及临床检查(主要... 抑郁症作为一类常见的、可治愈型的精神类疾病,若能在早期阶段对其进行有效筛查(即早期筛查)并及时采取相应的治疗手段,则可有效控制病情的进一步加重,甚至彻底治愈。传统的抑郁症诊断方法主要是医生通过患者的临床表现及临床检查(主要为诊断量表)进行综合判断,但诊断结果的准确与否严重依赖于医生的临床经验以及患者的高度配合。同时,由于抑郁症早期患者往往缺乏明显的病症表征,也极大增加了漏诊误诊的可能性。相关研究表明,脑电图(Electroencephalogram,EEG)能够反应受试者的精神状态,这为抑郁症的早期筛查提供了一种有效途径。基于此,以EEG信号为数据源,提出了一种基于EEG信号与深度学习的抑郁症早期筛查方法。首先,结合分段处理、频域转化等方法,对EEG信号进行时-频-空特征序列的提取;其次,基于所提特征序列与深度学习,构建了一种深度混合模型,通过训练模型完成正常人与轻度抑郁症患者的有效识别;最后,在公开数据集MODMA上验证所提方法的可行性与有效性。实验结果显示,早期筛查准确率为82.64%,召回率为78.42%,灵敏度为75.37%。 展开更多
关键词 抑郁症 脑电信号 早期筛查 --特征序列 深度混合模型
下载PDF
Spatial Distribution Characteristics of Surface Tidal Currents in the Southwest of Taiwan Strait 被引量:1
6
作者 SHEN Zhiben WU Xiongbin +3 位作者 LIN Hang CHEN Xiaofeng XU Xing’an LI Lun 《Journal of Ocean University of China》 SCIE CAS 2014年第6期971-978,共8页
This study was conducted on the spatial distribution characteristics of surface tidal currents in the southwestern Taiwan Strait based on the quasi-harmonic analysis of current data obtained by two high frequency surf... This study was conducted on the spatial distribution characteristics of surface tidal currents in the southwestern Taiwan Strait based on the quasi-harmonic analysis of current data obtained by two high frequency surface wave radar(HFSWR) systems. The analysis shows that the tidal current pattern in the southwestern Taiwan Strait is primarily semi-diurnal and influenced significantly by shallow water constituents. The spatial distribution of tidal current ellipses of M2 is probably affected by the interaction between two different systems of tide wave, one from the northern mouth of Taiwan Strait and the other from the Bashi Channel. The directions of the major axes of M2 tidal current ellipses coincide roughly with the axis of the Taiwan Strait. The spatial distribution of the magnitudes of the probable maximum current velocity(PMCS) shows gradual increase of the velocity from northeast to southwest, which is in accordance with the spatial distribution of the measured maximum current velocity(MMCS). The directions of the residual currents are in accordance with the direction of the prevailing monsoon wind at the Taiwan Strait and the direction of the Taiwan warm current during summer. The bathymetry also shows a significant effect on the spatial distribution characteristics of tidal currents. 展开更多
关键词 high frequency surface wave radar quasi-harmonic analysis spatial distribution characteristic surface tidal current Taiwan Strait
下载PDF
上一页 1 下一页 到第
使用帮助 返回顶部