Robust classification of EEG signal for brain-computer interface.

We report the implementation of a text input application (speller) based on the P300 event related potential. We obtain high accuracies by using an SVM classifier and a novel feature. These techniques enable us to maintain fast performance without sacrificing the accuracy, thus making the speller usable in an online mode. In order to further improve the usability, we perform various studies on the data with a view to minimizing the training time required. We present data collected from nine healthy subjects, along with the high accuracies (of the order of 95% or more) measured online. We show that the training time can be further reduced by a factor of two from its current value of about 20 min. High accuracy, fast learning, and online performance make this P300 speller a potential communication tool for severely disabled individuals, who have lost all other means of communication and are otherwise cut off from the world, provided their disability does not interfere with the performance of the speller.     

荧光免疫层析试条光电信号处理及特征量选取

荧光免疫层析定量检测技术是一种正在兴起的即时检验技术。在前期荧光免疫层析定量检测仪的研究基础上,采用小波算法对A/D采样后的荧光信号进行去噪处理,选用中值滤波剔除基线漂移,最后比较了采用测试线峰值和面积作为特征量的定量检测效果。数据显示,去噪后信噪比提高10 d B,相对均方差减小30.4%;采用峰值和面积值作为定量特征量的测量CV值分别为5.1%和0.4%,测量误差分别为9.1%和0.9%。结果表明采用积分面积作为特征量对荧光信号进行定量检测具有更好的准确度和重复性。     

微弱信号检测的3种非线性方法

阐述了随机共振法、混沌振子法及差分振子法3种方法的基本监测思想,确定了数学模型。随机共振系统(SR)是一个非线性双稳态系统,存在着在某一最佳输入噪声强度下,使系统产生最高信噪比输出,达到抑制噪声、放大微弱信号的目的。SR系统数学模型可由非线性Langevin方程定义,并进行了分析及仿真。混沌系统具有对初值敏感性及对噪声免疫的特点,数学模型选用Holmes型Duffing方程为监测器,进行了分析及仿真。差分振子法是基于差分方程构造监测器,以二维离散性系统作为数学模型,进行了分析及仿真。将3种方法应用于同步发电机转子匝间短路故障监测和异步电动机转子断条故障监测的实例中是成功的,证明了3种方法的有效性和准确性。     

Geometric subspace methods and time-delay embedding for EEG artifact removal and classification.

Generalized singular-value decomposition is used to separate multichannel electroencephalogram (EEG) into components found by optimizing a signal-to-noise quotient. These components are used to filter out artifacts. Short-time principal components analysis of time-delay embedded EEG is used to represent windowed EEG data to classify EEG according to which mental task is being performed. Examples are presented of the filtering of various artifacts and results are shown of classification of EEG from five mental tasks using committees of decision trees.     

基于多粒度特征选择和模型融合的复合电能质量扰动 分类特征优化

现代电力系统因其“双高”特性造成电能质量扰动模式愈加复杂,对复合扰动的准确分类提出了挑战。传统电能质量扰动分类方法在特征提取阶段所提取的特征由人为确定,难以判断所提取的特征对分类问题是否有效,加之多重复合扰动特征相互耦合导致扰动特征的可分性确定困难。为此,提出一种基于粒度的计算方法进行特征选择的模型。在提取的扰动特征集的基础上,通过构建多粒度空间反映特征分布差异性,进而挖掘各粒度下的最优特征子集以确定有效和冗余的分类特征,达到优化分类效果的目的。在此基础上,通过集成分类模型融合不同粒度空间最优扰动特征集所训练的同质弱分类器模型,提出一种新的电能质量扰动多粒度集成分类方法。该方法克服了现有方法在进行多粒度分类时通过寻找最优单粒度空间特征而导致的其他粒度空间信息丢失的问题。实验表明,多粒度特征选择算法可提取对分类有效的扰动特征,集成分类模型可进一步改善模型的分类性能。     

Wavelet-based feature extraction and selection for classification of power system disturbances using support vector machines

This paper presents a new approach for the classification of the power system disturbances using support vector machines (SVMs). The proposed approach is carried out at three serial stages. Firstly, the features to be form the SVM classifier are obtained by using the wavelet transform and a few different feature extraction techniques. Secondly, the features exposing the best classification accuracy of these features are selected by a feature selection technique called as sequential forward selection. Thirdly, the best appropriate input vector for SVM classifier is rummaged. The input vector is started with the first best feature and incrementally added the chosen features. After the addition of each feature, the performance of the SVM is evaluated. The kernel and penalty parameters of the SVM are determined by cross-validation. The parameter set that gives the smallest misclassification error is retained. Finally, both the noisy and noiseless signals are applied to the classifier given above stages. Experimental results indicate that the proposed classifier is robust and has more high classification accuracy with regard to the other approaches in the literature for this problem.     

Comparison of methods for forcing pulse linear electromagnetic motors

The method for forcing in the mode of the static inductive accumulator of magnetic energy is considered. For pulse linear electromagnetic motors used in impact devices, this method is shown to be advantageous as compared to the forcing method based on power supply from a source of increased voltage.     

基于BP神经网络的运动想象脑电信号分类研究

运动想象脑机接口因具有更大的自主性、灵活性,在脑机互联领域得到了广泛应用,相比较其它范式分类准确率偏低,限制了其发展。本文利用时频图谱、脑地形图两种特征分析方法对上肢运动想象脑电信号进行了特征分析,并采用滤波器组共空间模式(filter bank co-space,FBCSP)特征提取算法对上肢运动想象信号数据进行了特征提取,再将提取结果分别利用支持向量机（support vector machine,SVM）算法、K-最近邻(K-Nearest Neighbor)算法、反向传播（back propagation,BP）神经网络三种分类算法进行分类,研究结果发现SVM算法、KNN算法、BP神经网络算法应用在上肢运动想象脑机接口系统的平均分类准确率分别为76.45%、74.55%、81.70%,BP神经网络算法相比SVM算法、KNN算法在上肢运动想象任务的分类准确率上分别高出了5.25%、7.15%,并且t检验后得到分类准确率均具有极显著的统计学差异,并利用ROC曲线和AUC值检测了分类器效果,BP神经网络的AUC值相比SVM算法、KNN算法也分别提升了0.1226、0.1285,表明BP神经网络分类算法相比较SVM算法、KNN算法更适用于上肢运动想象脑机接口系统,提高了系统的分类准确率,推动了上肢运动想象脑电信号实际应用的发展进程。     

DOPO制备方法的比较与选择

综述了阻燃剂9，10-二氢-9氧杂-10-瞵杂菲-10-氧化物(DOPO)的制备过程，对每一步中间产品的制备方法做了详细的评述，着重对反应过程物料配比、反应温度、反应时间、体系压力等工艺参数以及提纯方法进行了比较、分析，提出了建议。     

Tracking EEG signal dynamics during mental tasks. A combinedlinear/nonlinear approach

The authors propose a combined approach using both linear and nonlinear methods to track the dynamics of short-term segments of EEG activity. This method was applied to EEG data collected during performance of voluntary movements under a condition of free choice of the instant of task performance for: (a) fine goal-directed movements to reach a chosen target, and (b) reproducing a temporal interval by twice pressing a button. In such an experimental set-up, the EEG activity is expected to change its characteristics in the short time periods passing from idle to active stage, the latter probably consisting of several successive phases. The hypothesis was that the dynamic state of EEG activity should change, and thus the EEG signal should be processed in segments with the use of different methods. Two linear methods were used: (a) singular spectrum analysis (SSA) to detect the temporal dynamic changes in decomposed segments of significant frequency bands, and (b) time-frequency analysis based on auto-regressive (AR) model coefficients, where abrupt short transients of their values were exhibited. To test whether linear techniques capture all of the information in the time series, four nonlinear methods were applied, as a function of time: point-wise dimension, Kolmogorov entropy, largest Lyapunov exponent, and nonlinear prediction. The results suggest that even in short periods, the EEG signal changes its dynamic structure, and thus different methods for dynamic analysis should be used     

脉搏信号中有效信号识别与特征提取方法研究

脉搏信号是重要的人体生理信号,但采集过程中会存在一定的干扰信号。针对人体脉搏波采集后的有效信号识别与特征提取的问题,提出一种基于时间序列描述的信号识别方法,首先将脉搏时间序列进行分割,每个分割段采用斜率符号化进行表示,通过段与段之间的相似性判断出信号的有用段和干扰段。再根据得到的有用段信号,提出一种基于滑窗的特征提取方法,寻找脉搏信号中的峰值、谷值,同时调整滑窗宽度,还能够进行重搏波波峰的检测。经实验验证,所提出的识别与特征提取方法准确率高且抗干扰性强。     

Optimal feature selection for classification of the power quality events using wavelet transform and least squares support vector machines

In this paper, a new optimal feature selection based power quality event recognition system is proposed for the classification of power quality events. While Apriori algorithm is capable of processing categorical data, an effective feature vector, which represents distinctive features of digital power quality event data, has been obtained by means of the proposed k-means based Apriori algorithm feature selection approach. The proposed k-means based Apriori algorithm feature selection approach is presented with a power quality event recognition system. In the power quality event recognition system, normalization and segmentation processes have been applied to three-phase event voltage signals. Using 9-level multiresolution analysis, wavelet transform coefficients of the event signals have been obtained. By applying nine different feature extraction processes to these coefficients, a 90 dimensional feature vector belonging to three-phase event voltage signals has been extracted. Optimal feature vector has been obtained by applying the k-means based Apriori algorithm feature selection approach to the obtained feature vector, which has been applied as the last step to the input of the least squares support vector machine classifier and recognition performance results have been obtained. Real power quality event data have been used to evaluate the performance of the proposed feature selection approach and power quality event recognition system. According to the results, the proposed k-means based Apriori algorithm feature selection approach and power quality event recognition system are efficient, reliable and applicable and classify three-phase event types with a high degree of accuracy.     

基于类心和特征加权的特征选择算法

基于边界最大化的特征选择方法是一种有效的特征选择方法,它能够显著去除高维数据中的不相干特征,在机器学习中有着重要的应用.但该方法存在着计算复杂度较大的问题,为了克服这一问题,提出了基于类心和特征加权的特征选择算法.其基本思想是以某一类的类心为中心,寻找其同类和异类最近邻构成边界,根据某种准则获得一个特征空间的权重,使得权重特征空间中的边界最大.在4个UCI数据库上的实验验证了所提算法不仅有更高的效率而且有更好的分类准确度,并且对于不相干特征几乎是不敏感的.     

旋转机械状态非线性特征提取及状态分类

为提取淹没于环境和结构噪声下风力机轴承故障信号,基于能量追踪法,提出改进变分模态分解法(improved variational mode decomposition, IVMD),并采用粒子群算法求解最优约束因子,获取准确模态分量;摒弃传统对故障特征频分量的提取,基于非线性分形理论提出多重分形谱特征因子(multi-fractal spectrum,MFC)以选取最具非线性特征的模态分量,以不同故障程度及状态的轴承加速度信号为对象,采用优化递归变分模态分解获取多分量,通过多重分形谱特征因子最大值选取有效信息分量,通过支持向量机进行故障分类。结果表明优化递归变分模态分解可准确分解振动信号至不同频段,以便故障信息提取;多重分形谱特征因子与信噪比呈正相关,以其最大值选取的分量具备更多有效信息;对IVMD-MFC所选取非线性分量,通过8种非线性特征值构建特征矩阵,通过BP神经网络实现故障分类,诊断准确度达97.5%。表明所提出方法可对不同故障程度的轴承状态进行区分。     

Using time-dependent neural networks for EEG classification.

This paper compares two different topologies of neural networks. They are used to classify single trial electroencephalograph (EEG) data from a brain-computer interface (BCI). A short introduction to time series classification is given, and the used classifiers are described. Standard multilayer perceptrons (MLPs) are used as a standard method for classification. They are compared to finite impulse response (FIR) MLPs, which use FIR filters instead of static weights to allow temporal processing inside the classifier. A theoretical comparison of the two architectures is presented. The results of a BCI experiment with three different subjects are given and discussed. These results demonstrate the higher performance of the FIR MLP compared with the standard MLP.     

基于 tSNE-ASC 特征选择和 DSmT 融合决策的 滚动轴承声振信号故障诊断

针对滚动轴承早期故障特征微弱且难以有效辨识的问题,提出一种基于tSNE-ASC特征选择和DSmT融合决策的滚动轴承声振信号故障诊断方法。利用多个传感器采集轴承在不同故障模式下的声振信号,将每个信号通过VMD分解得到K个IMF分量;对各个IMF分量进行特征提取,构建各个特征的数据集矩阵;利用tSNE将各特征数据集矩阵降维至二维,计算平均轮廓系数(ASC);根据ASC大于临界值提取出声振故障信号的敏感特征;基于诊断模型实现轴承故障的初级诊断;利用DSmT将声振信号初级诊断结果进行融合决策,得出最终的诊断结论。实验结果表明：基于tSNE-ASC的特征选择方法能有效提取混合域特征中的敏感特征,在不同工况、不同诊断模型中均具有很高的诊断精度;DSmT决策融合有效降低了单一信号诊断的不确定性,在变载荷和升降速非平稳工况下均有很高的诊断精度。     

微机保护中几种序分量选相方法的比较

分析比较了微机保护中几种常见序分量选相元件的性能,在此基础上提出了一种电压序分量的新型选相元件,具有一定的理论价值和实用价值。     

Comparison of HEMT non‐linear model extraction approaches based on small signal and on large signal measurements

In this paper the extraction of a non‐linear model for a HEMT is undertaken by means of two different approaches. The first approach is the classical procedure that consists of using the small signal data to fully generate a non‐linear model. The second alternative is developed following a different tactics that uses directly the response of the device to a large signal stimulus. Both approaches are compared in terms of required extraction measurements and limits of application. Copyright © 2002 John Wiley & Sons, Ltd.     

面向特征选择任务的改进蜣螂优化算法

蜣螂优化算法是一种基于蜣螂不同行为模式的新型启发式算法,与其他算法相比的收敛速度更快,逃脱局部最优的能力更强。针对蜣螂优化算法不能进行特征选择的问题,在蜣螂优化算法的基础上提出了蜣螂灰狼融合算法。该算法基于3种改进策略：精英初始化种群策略、灰狼蜣螂融合策略、运行加速策略,进一步提高蜣螂优化算法在特征选择任务上的性能,并给出了算法整体的伪代码。实验结果表明,比较其他改进型启发式算法,蜣螂灰狼融合优化算法在12个分类数据集中能够得到更高精度、更低维度的特征子集,同时兼备收敛速度、运行速度更快的优点。     

Single-trial EEG classification using in-phase average for brain-computer interface

Communication signals should be estimated by a single trial in a brain-computer interface. Since the relativity of visual evoked potentials from different sites should be stronger than those of the spontaneous electroencephalogram (EEG), this paper adopted the time-lock averaged signals from multi-channels as features. 200 trials of EEG recordings evoked by target or non-target stimuli were classified by the support vector machine (SVM). Results show that a classification accuracy of higher than 97% can be obtained by merely using the 250–550 ms time section of the averaged signals with channel Cz and Pz as features. It suggests that a possible approach to boost communication speed and simplify the designation of the brain-computer interface (BCI) system is worthy of an attempt in this way. __________ Translated from Journal of Huazhong University of Science and Technology (Nature Science Edition), 2007, 35(1): 11–13 [译自: 华中科技大学学报(自然科学版)]