EEG Feature Learning Model Based on Intrinsic Time-Scale Decomposition and Adaptive Huber Loss

According to the World Health Organization,about 50 million people worldwide suffer from epilepsy.The detection and treatment of epilepsy face great challenges.Electroencephalogram(EEG)is a significant research object widely used in diagnosis and treatment of epilepsy.In this paper,an adaptive feature learning model for EEG signals is proposed,which combines Huber loss function with adaptive weight penalty term.Firstly,each EEG signal is decomposed by intrinsic time-scale decomposition.Secondly,the statistical index values are calculated from the instantaneous amplitude and frequency of every component and fed into the proposed model.Finally,the discriminative features learned by the proposed model are used to detect seizures.Our main innovation is to consider a highly flexible penalization based on Huber loss function,which can set different weights according to the influence of different features on epilepsy detection.Besides,the new model can be solved by proximal alternating direction multiplier method,which can effectively ensure the convergence of the algorithm.The performance of the proposed method is evaluated on three public EEG datasets provided by the Bonn University,Childrens Hospital Boston-Massachusetts Institute of Technology,and Neurological and Sleep Center at Hauz Khas,New Delhi(New Delhi Epilepsy data).The recognition accuracy on these two datasets is 98%and 99.05%,respectively,indicating the application value of the new model.

基于Intrinsic Time-scale Decomposition算法的扰动信号特征量提取问题研究

由于电能质量扰动信号具有非线性的特点,当下算法无法有效提取电能质量扰动信号,采取Intrinsic Timescale Decomposition算法,可以有效提取电能质量扰动信号的频率、幅值、相位等特征量。利用Intrinsic Time-scale Decomposition算法提取电能质量扰动信号的Proper Rotation Component分量,后对Proper Rotation Component分量进行Hilbert变换以求得电能质量扰动信号的相位和瞬时频率,再根据电能质量扰动信号的高频突变点得到电能质量扰动信号的起止时间;根据envelope函数求得衰减因子。利用软件开发程序MATHEMATICA对电能质量扰动信号进行分析,结果表明,Intrinsic Time-scale Decomposition算法可以很好地提取电能质量扰动信号特征量,并且具有极佳的抗噪性能。

Real-time Subsynchronous Control Interaction Monitoring Using Improved Intrinsic Time-scale Decomposition

In recent years,subsynchronous control interaction(SSCI)has frequently taken place in renewable-connected power systems.To counter this issue,utilities have been seeking tools for fast and accurate identification of SSCI events.The main challenges of SSCI monitoring are the time-varying nature and uncertain modes of SSCI events.Accordingly,this paper presents a simple but effective method that takes advantage of intrinsic time-scale decomposition(ITD).The main purpose is to improve the accuracy and robustness of ITD by incorporating the least-squares method.Results show that the proposed method strikes a good balance between dynamic performance and estimation accuracy.More importantly,the method does not require any prior information,and its performance is therefore not affected by the frequency constitution of the SSCI.Comprehensive comparative studies are conducted to demonstrate the usefulness of the method through synthetic signals,electromagnetic temporary program(EMTP)simulations,and field-recorded SSCI data.Finally,real-time simulation tests are conducted to show the feasibility of the method for real-time monitoring.

Application of complete ensemble intrinsic time-scale decomposition and least-square SVM optimized using hybrid DE and PSO to fault diagnosis of diesel engines

针对固有时间尺度分解算法的模态混叠问题和最小二乘支持向量机的参数优化问题,本文提出了一种新的基于完备集合固有时间尺度分解和混合差分进化和粒子群算法优化最小二乘支持向量机的柴油机故障诊断方法。该方法主要包括以下几个步骤:首先,为解决固有时间尺度分解算法的模态混叠问题,提出了一种完备集合固有时间尺度分解算法。随后,利用完备集合固有时间尺度分解算法将非平稳的柴油机振动信号分解为一系列平稳的旋转分量和残差信号。然后,提取了前几阶旋转分量的三类典型的时频特征,包括奇异值、旋转分量能量和能量熵、AR模型参数,作为故障特征。最后,提出了混合差分进化和粒子群算法对最小二乘支持向量机的参数进行优化的方法,并通过将故障特征输入训练好的最小二乘支持向量机模型实现故障诊断。仿真和实验结果表明提出的故障诊断方法可以克服固有时间尺度分解的模态混叠问题,而且能够准确的识别柴油机故障。

Modal identification of multi-degree-of-freedom structures based on intrinsic chirp component decomposition method

Modal parameter identification is a mature technology.However,there are some challenges in its practical applications such as the identification of vibration systems involving closely spaced modes and intensive noise contamination.This paper proposes a new time-frequency method based on intrinsic chirp component decomposition(ICCD)to address these issues.In this method,a redundant Fourier model is used to ameliorate border distortions and improve the accuracy of signal reconstruction.The effectiveness and accuracy of the proposed method are illustrated using three examples:a cantilever beam structure with intensive noise contamination or environmental interference,a four-degree-of-freedom structure with two closely spaced modes,and an impact test on a cantilever rectangular plate.By comparison with the identification method based on the empirical wavelet transform(EWT),it is shown that the presented method is effective,even in a high-noise environment,and the dynamic characteristics of closely spaced modes are accurately determined.

Adaptive Variational Mode Decomposition for Bearing Fault Detection

Rolling element bearings are commonly used in rotary mechanical and electrical equipment. According to investigation, more than half of rotating machinery defects are related to bearing faults. However, reliable bearing fault detection still remains a challenging task, especially in industrial applications. The objective of this work is to propose an adaptive variational mode decomposition (AVMD) technique for non-stationary signal analysis and bearing fault detection. The AVMD includes several steps in processing: 1) Signal characteristics are analyzed to determine the signal center frequency and the related parameters. 2) The ensemble-kurtosis index is suggested to decompose the target signal and select the most representative intrinsic mode functions (IMFs). 3) The envelope spectrum analysis is performed using the selected IMFs to identify the characteristic features for bearing fault detection. The effectiveness of the proposed AVMD technique is examined by experimental tests under different bearing conditions, with the comparison of other related bearing fault techniques.

Regional features of the temperature trend in China based on Empirical Mode Decomposition

By the Empirical Mode Decomposition method, we analyzed the observed monthly average temperature in more than 700 stations from 1951-2001 over China. Simultaneously, the temperature variability of each station is calculated by this method, and classification chart of long term trend and temperature variability distributing chart of China are obtained, supported by GIS, 1 kmxl km resolution. The results show that： in recent 50 years, the temperature has increased by more than 0.4~C/10a in most parts of northern China, while in Southwest China and the middle and lower Yangtze Valley, the increase is not significant. The areas with a negative temperature change rate are distributed sporadically in Southwest China. Meanwhile, the temperature data from 1881 to 2001 in nine study regions in China are also analyzed, indicating that in the past 100 years, the temperature has been increasing all the way in Northeast China, North China, South China, Northwest China and Xinjiang and declining in Southwest China. An inverse ‘V-shaped’ trend is also found in Central China. But in Tibet the change is less significant.

Pressure fluctuation signal analysis of pump based on ensemble empirical mode decomposition method

Pressure fluctuations, which are inevitable in the operation of pumps, have a strong non-stationary characteristic and contain a great deal of important information representing the operation conditions. With an axial-flow pump as an example, a new method for time-frequency analysis based on the ensemble empirical mode decomposition （EEMD） method is proposed for research on the characteristics of pressure fluctuations. First, the pressure fluctuation signals are preprocessed with the empirical mode decomposition （EMD） method, and intrinsic mode functions （IMFs） are extracted. Second, the EEMD method is used to extract more precise decomposition results, and the number of iterations is determined according to the number of IMFs produced by the EMD method. Third, correlation coefficients between IMFs produced by the EMD and EEMD methods and the original signal are calculated, and the most sensitive IMFs are chosen to analyze the frequency spectrum. Finally, the operation conditions of the pump are identified with the frequency features. The results show that, compared with the EMD method, the EEMD method can improve the time-frequency resolution and extract main vibration components from pressure fluctuation signals.

GEARBOX FAULTDIAGNOSIS BASED ON EMPIRICAL MODE DECOMPOSITION

Time synchronous averaging of vibration data is a fundament technique forgearbox diagnosis. Currently, this technique relies on hardware tachometer to give phase synchronousinformation. Empirical mode decomposition (EMD) is introduced to replace time synchronous averagingof gearbox vibration signal. With it, any complicated dataset can be decomposed into a finite andoften small number of intrinsic mode functions (IMF). The key problem is how to assure thatvibration signals deduced by gear defects could be sifted out by EMD. The characteristic vibrationsignals of gear defects are proved IMFs, which makes it possible to utilize EMD for the diagnosis ofgearbox faults. The method is validated by data from recordings of the vibration of a single-stagespiral bevel gearbox with fatigue pitting. The results show EMD is powerful to extractcharacteristic information from noisy vibration signals.

A method for extracting human gait series from accelerometer signals based on the ensemble empirical mode decomposition

In this paper, the ensemble empirical mode decomposition （EEMD） is applied to analyse accelerometer signals collected during normal human walking. First, the self-adaptive feature of EEMD is utilised to decompose the ac- celerometer signals, thus sifting out several intrinsic mode functions （IMFs） at disparate scales. Then, gait series can be extracted through peak detection from the eigen IMF that best represents gait rhythmicity. Compared with the method based on the empirical mode decomposition （EMD）, the EEMD-based method has the following advantages： it remarkably improves the detection rate of peak values hidden in the original accelerometer signal, even when the signal is severely contaminated by the intermittent noises; this method effectively prevents the phenomenon of mode mixing found in the process of EMD. And a reasonable selection of parameters for the stop-filtering criteria can improve the calculation speed of the EEMD-based method. Meanwhile, the endpoint effect can be suppressed by using the auto regressive and moving average model to extend a short-time series in dual directions. The results suggest that EEMD is a powerful tool for extraction of gait rhythmicity and it also provides valuable clues for extracting eigen rhythm of other physiological signals.

Anchoring Bolt Detection Based on Morphological Filtering and Variational Modal Decomposition

The pull test is a damaging detection method that fails to measure the actual length of a bolt.Thus,the ultrasonic echo is an important non?destructive testing method for bolt quality detection.In this research,the variational modal decomposition(VMD)method is introduced into the bolt detection signal analysis.On the basis of morphological filtering(MF)and the VMD method,a VMD?combined MF principle is established into a bolt detection signal analysis method(MF?VMD).MF?VMD is used to analyze the vibration and actual bolt detection signals of the simulation.Results show that MF?VMD effectively separates intrinsic mode function,even under strong interference.In comparison with conventional VMD method,the proposed method can remove noise interference.An intrinsic mode function of the field detection signal can be effectively identified by reflecting the signal at the bottom of the bolt.

Improvement of the Mirror Extending in Empirical Mode Decomposition Method and the Technology for Eliminating Frequency Mixing

The mirror extending approach proposed by Zhao and Huang in EMD method is improved in this paper. Mirror extending manner of data is kept unchanged, but the approach for determining envelopes is changed. When the end of data is obviously not extremum, the envelope is determined by the first inner extremum and the image value in the mirror, ignoring the value on the end. This improvement eliminates the frequency compression near the end and decreases the error. Meanwhile, tridiagonal equations are used and the calculation speed is much increased. The temporal process curve is more important in reflecting the real physical process and comparable with other phenomena. Frequency mixing in IMFs makes it impossible. A high frequency reconstruction (HFR) approach is proposed to eliminate common frequency mixing and reconstruct an IMF with all high frequency portions. By this approach, the IMFs without frequency mixing are obtained to express significative processes. The high frequency information restored in high frequency IMF can be extracted by general spectrum method. After obtaining IMFs by EMD method, some of the theoretical and technological issues still exist when using the IMFs. The consistency of IMFs with real physical process is discussed in detail. By virtue of the approach proposed in this paper, the EMD method can be widely used in various fields.

Multi-scale spatial relationships between soil total nitrogen and influencing factors in a basin landscape based on multivariate empirical mode decomposition

The relationships between soil total nitrogen(STN)and influencing factors are scale-dependent.The objective of this study was to identify the multi-scale spatial relationships of STN with selected environmental factors(elevation,slope and topographic wetness index),intrinsic soil factors(soil bulk density,sand content,silt content,and clay content)and combined environmental factors(including the first two principal components(PC1 and PC2)of the Vis-NIR soil spectra)along three sampling transects located at the upstream,midstream and downstream of Taiyuan Basin on the Chinese Loess Plateau.We separated the multivariate data series of STN and influencing factors at each transect into six intrinsic mode functions(IMFs)and one residue by multivariate empirical mode decomposition(MEMD).Meanwhile,we obtained the predicted equations of STN based on MEMD by stepwise multiple linear regression(SMLR).The results indicated that the dominant scales of explained variance in STN were at scale 995 m for transect 1,at scales 956 and 8852 m for transect 2,and at scales 972,5716 and 12,317 m for transect 3.Multi-scale correlation coefficients between STN and influencing factors were less significant in transect 3 than in transects 1 and 2.The goodness of fit root mean square error(RMSE),normalized root mean square error(NRMSE),and coefficient of determination(R2)indicated that the prediction of STN at the sampling scale by summing all of the predicted IMFs and residue was more accurate than that by SMLR directly.Therefore,the multi-scale method of MEMD has a good potential in characterizing the multi-scale spatial relationships between STN and influencing factors at the basin landscape scale.

Segmented second algorithm of empirical mode decomposition

A new algorithm, named segmented second empirical mode decomposition （EMD） algorithm, is proposed in this paper in order to reduce the computing time of EMD and make EMD algorithm available to online time-frequency analysis. The original data is divided into some segments with the same length. Each segment data is processed based on the principle of the first-level EMD decomposition. The algorithm is compared with the traditional EMD and results show that it is more useful and effective for analyzing nonlinear and non-stationary signals.

Weighing axle weight of moving vehicle based on empirical mode decomposition

Dynanfic forces are the main factor that influences the axle weight measurement accuracy of moving vehicle. Empirical mode decomposition （EMD） is presented to separate the dynamic forces contained in the axle weight signal. The concept and algorithm of EMD are introduced. The characteristic of the axle weight signal is analyzed. The method of judging pseudo intrinsic mode function （pseudo-IMF） is presented to improve the weighing accuracy. Numerical simulation and field experiments are conducted to evaluate the performance of EMD. The result shows effectiveness of the proposed method. Maximum weighing errors of the front axle, the rear axle and the gross weight at the speed of 15 km/h or lower are 2.22%, 6.26% and 4.11% respectively.

Signal prediction based on empirical mode decomposition and artificial neural networks

In view of the usefulness of Empirical Mode Decomposition （EMD）, Artificial Neural Networks （ ANN）, and Most Relevant Matching Extension （MRME） methods in dealing with nonlinear signals, we pro- pose a new way of combining these methods to deal with signal prediction. We found the results of combining EMD with either ANN or MRME to have higher prediction precision for a time series than the result of using EMD alone.

Discriminative feature encoding for intrinsic image decomposition

Intrinsic image decomposition is an important and long-standing computer vision problem.Given an input image,recovering the physical scene properties is ill-posed.Several physically motivated priors have been used to restrict the solution space of the optimization problem for intrinsic image decomposition.This work takes advantage of deep learning,and shows that it can solve this challenging computer vision problem with high efficiency.The focus lies in the feature encoding phase to extract discriminative features for different intrinsic layers from an input image.To achieve this goal,we explore the distinctive characteristics of different intrinsic components in the high-dimensional feature embedding space.We define feature distribution divergence to efficiently separate the feature vectors of different intrinsic components.The feature distributions are also constrained to fit the real ones through a feature distribution consistency.In addition,a data refinement approach is provided to remove data inconsistency from the Sintel dataset,making it more suitable for intrinsic image decomposition.Our method is also extended to intrinsic video decomposition based on pixel-wise correspondences between adjacent frames.Experimental results indicate that our proposed network structure can outperform the existing state-of-the-art.

一种灰色关联分析优化ICEEMDAN的VP倾斜仪信号降噪模型

VP倾斜仪固体潮信号受仪器监测复杂环境限制,多含有大量环境噪声。为获得真实固体潮曲线,提出一种基于灰色关联分析优化改进的自适应噪声完备集合经验模态分解(ICEEMDAN)VP倾斜仪信号降噪模型(GRA-ICEEMDAN)。该方法首先将含噪信号进行ICCEMDAN处理,得到若干个固有模态函数(IMF),并依次排列与标记;然后基于这些IMF分别计算相关系数、互信息、R^(2)、Adj-R^(2)、MSE、SSE、RMSE、MAE、MAPE、样本熵等10个评价指标值,构建IMF可信度评价指标矩阵;最后借助灰色关联分析(GRA)计算各评价指标与不同IMF之间的关联系数和关联度,依据关联度大小对各个IMF进行排序,将排名靠前的IMF进行线性重构,即可完成信号降噪。仿真去噪实验和实测去噪实验均表明,GRA-ICEEMDAN模型优于卡尔曼滤波、70阶低通FIR滤波、Savitzky-Golay等经典降噪模型,能显著区分噪声成分和有效成分,原始信号分解后的重构误差与信号损失极小,可推广至其他仪器的复杂信号降噪中。

基于参数优化VMD-小波阈值的轴承振动信号降噪方法

为了解决复杂工况下滚动轴承振动信号存在随机噪声的问题,提出了一种基于参数优化变分模态分解(VMD)-小波阈值的滚动轴承降噪方法。首先,利用以包络熵为适应度函数的天鹰算法对变分模态分解算法的模态分解数K和惩罚因子α进行了自适应选择,代入VMD分解中,得到若干本征模态函数(IMFs);然后,根据峭度-相关系数将IMF分量划分为纯净分量和含噪分量,对含噪分量进行了小波阈值降噪处理;最后,对处理后的分量进行了重构,并用重构信号进行了包络谱分析,实现了滚动轴承的信号降噪目的,并利用仿真信号和美国凯斯西储大学公开的轴承数据集对上述降噪方法的有效性进行了验证。研究结果表明:基于参数优化VMD-小波阈值的降噪方法减少了滚动轴承运行状态下的随机噪声,相对小波阈值降噪方法,所得仿真信号信噪比提升53%,均方误差降低13%;在故障特征频率为162 Hz时,所得实验降噪信号包络谱的前6倍频谱峰值更为明显,且受随机噪声影响较小。该研究方法在滚动轴承等旋转机械信号降噪方面具有一定的参考价值。

基于多域信息融合与深度分离卷积的轴承故障诊断网络模型

针对传统卷积神经网络(CNN)对滚动轴承振动信号的故障识别准确率不高这一问题,提出了一种基于多域信息融合结合深度分离卷积(MDIDSC)的轴承故障诊断方法。首先,利用自适应噪声的完全集合经验模态分解(CEEMDAN)算法对轴承振动信号进行了分解;然后,利用分解出的本征模态函数(IMF)的各个分量构建了多空间状态矩阵,并将该多空间状态矩阵输入该深度分离卷积模型中,进行了卷积训练;同时,在该深度分离卷积模型中添加了残差结构,对数据特征进行了复利用,并对卷积核进行了深度分离,解决了深度模型的网络退化问题;最后,提出了一种空间特征提取方法,对模型参数进行了修剪,采用一种自适应学习率退火方法进行了梯度优化,以避免模型陷入局部最优。研究结果表明:通过对多个轴承故障数据集进行对比分析可知,MDIDSC在轴承故障诊断方面的准确率和稳定性明显优于其他方法,MDIDSC的最高测试准确率为100%,平均测试准确率为99.07%;同时,在测试集中的最大损失和平均损失分别为0.1345和0.0841;该结果表明MDIDSC在轴承故障诊断方面具有一定的优越性。