基于ITD和PLV的四类运动想象脑电分类方法研究

针对四类运动想象任务的特征提取问题,提出一种基于固有时间尺度分解(ITD)和相位同步分析相结合的脑电(EEG)信号特征提取方法。采用第3届和第4届BCI竞赛中的4类运动想象数据集,首先选择5个导联的运动想象脑电信号,根据相位同步性计算导联之间的相锁值(PLV),将相锁值作为一类特征;之后利用ITD对5个导联的运动想象脑电信号进行分解,提取第一层固有旋转分量的能量特征,与PLV特征相结合获得十五维特征向量;最后通过支持向量机(SVM)进行分类识别。对12名受试者的平均识别率达到91. 64%,平均Kappa系数达到0. 887,说明该方法能够有效的提取脑电信号特征,进而提高4类运动想象任务的分类准确率。     

基于SSVEP信号的下肢外骨骼机器人控制系统研究

为了降低脑控下肢外骨骼机器人的研发成本,促进脑机接口技术的快速发展,提出一种基于深度学习与Matlab/Simulink联合仿真控制方法。该方法建立具有多个自由度的下肢外骨骼机器人样机模型,并进行运动学仿真,验证模型的合理性。利用EEGLAB对SSVEP信号进行滤波预处理,通过FFT变换将信号从时域转换到频域,提取SSVEP信号的频域特征。结合深度学习理论对特征进行分类,将分类结果转换成控制指令,控制仿真模型进行运动。实验结果表明,该方法控制的平均准确率达到了79.8%,验证了其可行性。     

基于多特征卷积神经网路的运动想象脑电信号分析及意图识别

为了准确提取个体运动想象脑电信号的最优时段和频段特征以及有效提高其分类准确率,结合卷积神经网络和集成分类方法提出一种多特征卷积神经网络(MFCNN)算法,对运动想象脑电信号进行分类识别。首先对脑电信号进行预处理,然后将原始信号、能量特征、功率谱特征以及融合特征分别输入到卷积神经网络中得到其训练模型,最后通过加权投票的集成分类方法得到最终的分类结果。并利用2008年BCI竞赛Datasets 2b数据集和实测数据对所提出的方法进行实验分析。结果表明,所提的MFCNN方法可有效提高运动想象识别率,实验中所有受试者的平均分类正确率和平均Kappa值分别为78.6%和0.57,为运动想象类脑机接口的应用提供了新的思路和方法。     

基于可解释深度学习的单通道脑电跨被试疲劳驾驶检测

脑电信号被认为是检测驾驶员疲劳状态的最佳生理信号之一。 然而,由于不同被试者和不同记录时段的脑电信号差异 很大,设计一个无校准的脑电疲劳检测系统仍然具有挑战性。 近年来,虽然开发了许多深度学习方法来解决这个问题并取得了 重大进展,但是深度学习模型的黑盒效应使得模型决策不可信赖。 为此,本文提出了一种可解释深度学习模型,用于从单通道 脑电信号中检测跨被试疲劳状态。 该模型具有紧凑的网络结构,首先设计浅层 CNN 提取 EEG 特征,然后引入自适应特征重新 校准机制增强提取特征的质量,最后通过 LSTM 网络将时间特征序列与分类相关联。 模型分类决策的可解释信息则是由 LSTM 输出隐藏状态的可视化技术实现的。 在持续驾驶任务的公开脑电数据集上进行大量跨被试实验,该模型的分类平均准确率最 高达到 76. 26% 。 相比于先进的紧凑型深度学习模型,该模型有效降低了参数量和计算量。 可视化结果表明该模型已发现神经 生理学上可靠的解释。     

基于免疫优化的脑电自适应集成分类方法研究

针对脑机接口研究(BCI)中对脑电波信号的分类识别问题,对脑电信号中P300脑电信号的预处理、特征提取及特征分类等方面算法进行了研究,主要侧重于对P300脑电信号分类算法的研究。提出了一种自适应的集成支持向量机(SVM)分类方法,利用免疫算法的多样性以及自我调节能力,对基于Bagging的集成SVM分类学习器进行了优化,提高了对P300脑电信号识别的准确度以及针对不同个体的自适应性。研究结果表明:将自适应集成分类算法运用在BCI Competition III Dataset II的P300脑电数据上,可以识别出被试者的脑电意图,并且对P300脑电信号的分类可以达到较高的分类准确率,实验结果稳定在98%。     

基于 HFD 和 LZC 特征联合的单通道 静息态脑电抑郁症识别研究

目前抑郁症的临床诊断多以医生经验和患者主观感受为主,主观性强、准确率低、耗时长。 随着神经电生理学和计算机 技术的发展,抑郁症的客观分类与识别成为可能。 但是,已有的基于静息态脑电信号的抑郁症分类识别方法较为单一,脑电特 征选取的精准性、综合性和有效性有待进一步探究。 本文在设计包含两种模态实验范式的基础上,提出一种基于 HFD 和 LZC 特征联合的单通道静息态脑电抑郁症分类识别方法,以期用较少的特征获得较高的分类准确率。 首先采集 8 名抑郁患者和 8 名健康对照的静息态脑电信号;然后提取其非线性动力学特征参数 HFD 和 LZC;最后将特征数据输入到非线性支持向量机模 型中进行分类识别。 结果表明,联合特征得到的灵敏度、特异性和分类正确率最高分别为 98. 12% 、96. 67% 和 95. 10% ,较单独 HFD/ LZC 特征平均分别提高了 23. 05% 、17. 02% 和 19. 29% 。 同时,模型主体部分仅耗时约 12 s。 研究结果对临床实践中抑郁 症的识别和辅助诊断具有重要意义。     

基于深度网络迁移学习的致痫区脑电识别* .txt

摘要：致痫区脑电识别能够为癫痫外科手术提供重要的参考价值。提出了一种基于深度网络迁移学习的致痫区脑电识别算法。首先利用连续小波变换（CWT）对脑电信号进行时频分析，获得脑电信号时频图；然后迁移学习AlexNet网络模型，调整网络结构使之适应于致痫区脑电识别，将模型第7层全连接层输出作为脑电信号时频图的特征表示，最后利用支持向量机（SVM）、BP神经网络、长短期记忆网络（LSTM）、基于稀疏表达分类算法（SRC）、线性判别分析（LDA）等分类算法进行特征分类。基于开源脑电数据集采用十折交叉验证的方法对算法进行了验证,比较6种分类器的效果，得到SVM算法的平均特异性为8881%，灵敏度为8807%，准确率为8844%，证明了该方法识别致痫区脑电信号的有效性。 .txt     

一种面向运动解码的 EEG-fNIRS 时频特征融合 与协同分类方法

脑功能成像技术可以反映人体运动时的大脑生理变化,进而解码运动状态,但单模态信号反映的大脑生理信息存在局 限性。 为此,本文提出了一种基于 EEG 和 fNIRS 信号的时频特征融合与协同分类方法,利用脑神经电活动和血氧信息的互补 特性提高运动状态解码精度。 首先,提取 EEG 的小波包能量熵特征,使用双向长短期记忆网络(Bi-LSTM)提取 fNIRS 的时域特 征,将两类特征组合得到包含时频域信息的融合特征,实现 EEG 和 fNIRS 不同层次特征的信息互补。 然后,利用 1DCNN 提取 融合特征深层次信息。 最后,采用全连接神经网络进行任务分类。 将所提方法应用于公开数据集,本文所提的 EEG-fNIRS 信号 协同分类方法准确率为 95. 31% ,较单模态分类高 7. 81% ~ 9. 60% 。 结果表明,该方法充分融合了两互补信号的时频域信息,提 高了对左右手握力运动的分类准确率。     

基于频带能量归一化和SVM-RFE的ECoG分类

针对基于运动想象(左手小手指和舌头)的皮层脑电(electrocortieographic,ECoG)信号的分类问题,对BC12005竞赛数据集Ⅰ中的ECoG信号使用频带能量(band power,BP)归一化算法提取运动相关电位(movement related potential,MRP)、μ节律和β节律的频带能量作为特征.针对特征提取后维数较高的问题,使用基于支持向量机的回归特征消去(support vector machinerecursive feature elimination,SVM-RFE)算法进行特征选择,通过对训练数据集使用lO段交叉验证(cross validation,CV)的方法寻找最佳特征组合,确定特征在维数为6时具有最低平均识别错误率,对测试数据集采用同样的方法和同样的组合进行特征提取,并使用线性支持向量机进行分类,分类正确率可以达到93%.     

基于贝叶斯分类的核电站泵类设备振动故障诊断方法

传统方法在诊断设备振动故障时,仅提取了振动信号时域特征作为故障向量,导致故障诊断准确率较低、诊断时间较长,故设计了基于贝叶斯分类的核电站泵类设备振动故障诊断方法,在采集泵类设备振动信号后,重构振动信号低频部分并提取信号的时域和频域向量。将提取结果作为贝叶斯分类器的条件属性变量,计算变量归于故障类别的信息熵,选择最高信息熵对应的故障类别作为诊断结果。结果表明：该方法在提高故障诊断准确率的同时缩短了诊断时间。     

A flexible classification method for evaluating the utility of automated workpiece classification system

In group technology, workpieces are categorised into families according to their similarity in design or manufacturing attributes. This categorisation can eliminate design duplication and facilitate the production of workpieces. Much effort has been focused on the development of automated workpiece classification systems. However, it is difficult to evaluate the utility of such systems. The objective of this study was to develop a benchmark classification system based on global shape information for use in evaluating the utility of workpiece classification systems. A classification system has a high level of utility if its classification scheme is consistent with users' perceptual judgment of the similarity between workpiece shapes. Hence, in the proposed method, the consistency between a classification system and users' perceptual judgements is used as an index of the utility of the system. The proposed benchmark classification has two salient characteristics:

Materials classification for fine mechanics

Materials selection processes consider not only the basic properties of materials but also products and ratios of groups of properties. This paper explores which groups might be used generally as the basic for classifying materials according to their performance in different application areas of precision mechanics and discusses how the data can be efficiently presented. The approach is illustrated by considering a selection of materials often used in precision designs.     

Basing-element classification for lathe attachments

A classification system is outlined for the factors that influence the choice of basing elements.     

Fault classification using genetic programming

Genetic programming (GP) is a stochastic process for automatically generating computer programs. In this paper, three GP-based approaches for solving multi-class classification problems in roller bearing fault detection are proposed. Single-GP maps all the classes onto the one-dimensional GP output. Independent-GPs singles out each class separately by evolving a binary GP for each class independently. Bundled-GPs also has one binary GP for each class, but these GPs are evolved together with the aim of selecting as few features as possible. The classification results and the features each algorithm has selected are compared with genetic algorithm (GA) based approaches GA/ANN and GA/SVM. Experiments show that bundled-GPs is strong in feature selection while retaining high performance, which equals or outperforms the two previous GA-based approaches.     

奇异点和隐马尔可夫模型融合的指纹分类

为了提高分类精度,提出一种基于奇异点和隐马尔可夫模型(HMM)融合的指纹分类方法.分别对基于奇异点的指纹分类方法和基于HMM的指纹分类方法的信任度函数进行分配,利用证据理论求得两种方法联合作用下的基本可信度分配值.最后,根据纹形模式判定规则.选择具有最大支持度的目标完成指纹纹型分类.利用提出的方法在国际指纹竞赛数据库上做了测试,总的纹型辨识平均正确率可达94.5%,识别结果优于奇异点分类方法和HMM分类方法,具有一定的实用价值.     

Machine-fault classification: A fuzzy-set approach

This is an experimental study to investigate the feasibility of employing fuzzysets theory in an integrated machine-fault diagnostic system. This system consists of modules for data acquisition, data processing, features extraction, fault clustering and fault assignment. Among these tasks, fault clustering and fault assignment are accomplished using fuzzy-sets-based procedures. The theoretical foundation of this study is discussed, with case studies.     

逆向物流分级化及集对分级技术的应用

针对国内逆向物流对策少,难度大,地区发展不均衡的特点,提出了逆向物流分级化的新思路。同时,借助集对分析理论,结合熵权系数,以区域分级为例建立了分级评价模型,为该对策的实施提供了科学依据。     

Fractal-wavelet based classification of tribological surfaces

Classification of the topography of freshly machined, worn and damaged surfaces (e.g. damaged by adhesion, scoring, abrasion, pitting) is still a problem in machine failure analysis. Tribological surfaces often exhibit both a multiscale nature (i.e. different length scales of surface features) and a non-stationary nature (i.e. features which are superimposed on each other and located at different positions on a surface). The most widely used approaches to surface classification are based on the Fourier transform or statistical functions and parameters. Often these approaches are inadequate and provide incorrect classification of the tribological surfaces. The main reason is that these techniques fail to simultaneously capture the multiscale nature and the non-stationary nature of the surface data. A new method, called a hybrid fractal-wavelet method, has recently been developed for the characterization of tribological surfaces in a multiscale and non-stationary manner. In contrast to other methods, this method combines both the wavelets’ inherent ability to characterize surfaces at each individual scale and the fractals’ inherent ability to characterize surfaces in a scale-invariant manner. The application of this method to the classification of artificially generated fractal and tribological surfaces (e.g. worn surfaces) is presented in this paper. The newly developed method has been further modified to better suit tribological surface data, including a new measure of differences between initial and decoded images. The accuracy of this method in the classification of surfaces was assessed.     

Surface roughness classification using image processing

Surface roughness is an important factor in determining the satisfactory functioning of the machined components. Conventionally the surface roughness measurement is done with a stylus instrument. Since this measurement process is intrusive and is of contact type, it is not suitable for online measurements. There is a growing need for a reliable, online and non-contact method for surface measurements. Over the last few years, advances in image processing techniques have provided a basis for developing image-based surface roughness measuring techniques. Based upon the vision system, novel methods used for human identification in biometrics are used in the present work for characterization of machined surfaces. The Euclidean and Hamming distances of the surface images are used for surface recognition. Using a CCD camera and polychromatic light source, low-incident-angle images of machined surfaces with different surface roughness values were captured. A signal vector was generated from image pixel intensity and was processed using MATLAB software. A database of reference images with known surface roughness values was established. The Euclidean and Hamming distances between any new test surface and the reference images in the database were used to predict the surface roughness of the test surface.