Novel Face Recognition Method by Combining Spatial Domain and Selected Complex Wavelet Features

A novel face recognition method based on fusion of spatial and frequency features was presented to improve recognition accuracy. Dual-Tree Complex Wavelet Transform derives desirable facial features to cope with the variation due to the illumination and facial expression changes. By adopting spectral regression and complex fusion technologies respectively, two improved neighborhood preserving discriminant analysis feature extraction methods were proposed to capture the face manifold structures and locality discriminatory information. Extensive experiments have been made to compare the recognition performance of the proposed method with some popular dimensionality reduction methods on ORL and Yale face databases. The results verify the effectiveness of the proposed method.

VARIABLE SELECTION BY PSEUDO WAVELETS IN HETEROSCEDASTIC REGRESSION MODELS INVOLVING TIME SERIES

A simple but efficient method has been proposed to select variables in heteroscedastic regression models. It is shown that the pseudo empirical wavelet coefficients corresponding to the significant explanatory variables in the regression models are clearly larger than those nonsignificant ones, on the basis of which a procedure is developed to select variables in regression models. The coefficients of the models are also estimated. All estimators are proved to be consistent.

Wavelet packet feature selection for lung sounds based on optimization

In this paper, a wavelet packet feature selection method for lung sounds based on optimization is proposed to obtain the best feature set which maximizes the differences between normal lung sounds and abnormal lung sounds(sounds with wheezes or rales). The proposed method includes two main steps: Firstly, the wavelet packet transform(WPT) is used to extract the original features of lung sounds; then the genetic algorithm(GA) is used to select the best feature set. The obtained optimal feature set is sent to four different classifiers to evaluate the performance of the proposed method. Experimental results show that the feature set obtained by the proposed method provides a higher classification accuracy of 94.6% in comparison with the best wavelet packet basis approach and multi-scale principal component analysis(PCA) approach. Meanwhile, the proposed method has effective generalization performance and can obtain the best feature set without priori knowledge of lung sounds.

小波DehazeFormer网络的道路交通图像去雾

针对道路交通雾图像对比度低、细节丢失、模糊和失真的问题,提出了一种小波DehazeFormer模型的道路交通图像去雾方法。为提升模型去雾能力,构建了编解码结构的小波DehazeFormer网络,编码器以DehazeFormer与选择性核特征融合模块(Selective kernel feature fusion,SKFF)级联作为骨干网络的基本单元,编码部分由三级这样的基本单元构成,以融合图像的原始信息和去雾后的信息,更好地捕获雾图中关键特征;中间特征层采用局部残差结构,并加入卷积注意力机制(Convolutional Block Attention Module,CBAM),对不同级别的特征赋予不同权重,同时融入内容引导注意力混合方案(Content-guided Attention based Mixup Fusion Scheme,CGAFusion),通过学习空间权重来调整特征;解码部分由DehazeFormer和SKFF构成,采用逐点卷积,在保证网络性能同时,减少网络的参数量;跳跃连接引入小波变换,对不同尺度的特征图进行小波分析,获取不同尺度的高、低频特征,放大交通雾图的细节使得复原图像保留纹理;最后,将原始图像和经解码后输出的特征图融合,获取更多的细节信息。实验结果表明,本文方法相比于基线DehazeFormer网络,其PSNR指标在公开数据集中提升1.32以上,在合成数据集中提升0.56,SSIM指标提升了0.015以上,MSE值有较大幅度降低,下降了23.15以上;Entropy指标提升0.06以上。本文去雾算法对提升交通雾图像的对比度、降低雾图模糊和失真及细节丢失等方面均表现出优良的性能,有助于后续道路交通的智能视觉监控与管理。

融合稳态和暂态特征量的接地故障选线方法研究

单相接地故障是配电网运行时发生概率最高的故障,但接地时存在电气故障特征弱、外界干扰大的情况,使得配电网存在接地选线困难的问题。对此,提出基于稳态和暂态故障特征量相融合的配电网接地选线方法,分析单相接地时接地故障线路与非故障线路对地电容电流突变量五次谐波分量在幅值和相位上的差异性,利用经验小波变换提取故障零序电流的低频分量综合相关系数和高频分量相对权重系数,并利用模糊神经网络实现融合特征量与故障的非线性映射诊断。建立配电网接地故障仿真模型,通过多重干扰下的选线对比分析,验证了本文方法的有效性和优越性。

Comparative Analysis of Wavelet Transform for Time-Frequency Analysis and Transient Localization in Structural Health Monitoring

A critical problem facing data collection in structural health monitoring,for instance via sensor networks,is how to extract the main components and useful features for damage detection.A structural dynamic measurement is more often a complex time-varying process and therefore,is prone to dynamic changes in time-frequency contents.To extract the signal components and capture the useful features associated with damage from such nonstationary signals,a technique that combines the time and frequency analysis and shows the signal evolution in both time and frequency is required.Wavelet analyses have proven to be a viable and effective tool in this regard.Wavelet transform(WT)can analyze different signal components and then comparing the characteristics of each signal with a resolution matched to its scale.However,the challenge is the selection of a proper wavelet since various wavelets with varied properties that are to analyze the same data may result in different results.This article presents a study on how to carry out a comparative analysis based on analytic wavelet scalograms,using structural dynamic acceleration responses,to evaluate the effectiveness of various wavelets for damage detection in civil structures.The scalogram’s informative time-frequency regions are examined to analyze the variation of wavelet coefficients and show how the frequency content of a signal changes over time to detect transient events due to damage.Subsequently,damage-induced changes are tracked with time-frequency representations.Towards this aim,energy distribution and sharing information are investigated.The undamaged and damaged simulated comparative results of a structure reveal that the damaged structure were shifted from the undamaged structure.Also,the Bump wavelet shows the best results than the others.

基于CWT-CNN-LSTM的配电网单相接地故障选线方法分析

阐述配电网单相接地故障特征提取难点,分析现有选线方法、选线精度不高的问题,提出一种连续小波变换(Continuous wavelet transform,CWT)和CNN-LSTM的故障选线方法。首先对零序暂态电流进行连续小波变换获取对应的时频灰度图像,然后CNN自适应提取时频灰度图像的局部特征,LSTM层从CNN层学到的局部特征中学习上下文依赖关系,最后通过SoftMax层实现故障选线。仿真结果表明,所提方法的选线精度为99.65%,与CWT-CNN等方法相比,具有较强的鲁棒性。

基于小波包分解多信息融合的配电网故障选线保护方法

为了解决配电网混合线路频发单相接地故障导致选线困难的问题,提出了一种新的保护方法。该方法通过db10小波包对配电网每条出线的零序电流进行第5层分解得到零序电流小波低频重构系数,运用Hausdorff距离算法将每条出线的低频重构系数进行计算得到不匹配度特征值;再结合小波包能量特点得到每条出线归一化后的综合小波能量特征值,将两种故障特征量作为输入量,结合随机森林算法具有数据融合和无需设定阀值的特性,建立配电网故障选线判别模型。通过将不同工况下得到的256组训练样本数据对模型进行训练获得判别模型的最优参数,再将剩余不同的48组测试数据进行模型验证,结果表明该方法在不同的故障条件下故障判别准确率高,具有较强的适用性。同时将所提出的方法与BP神经网络、ELM信息融合选线方法进行了对比,仿真结果表明,在加入信噪比为40 dB和20 dB的高斯白噪声的工况下,所提出的故障选线方法无论是分类准确率还是收敛时间都具有显著的优势。

基于最小二乘支持向量机的造纸工控网络高隐蔽性入侵检测

造纸工控网络的数据特征具有复杂性和多样性,对于高隐蔽性入侵行为,其特征可能被混杂在正常操作和噪声中,增加了检测的难度。为此,提出基于最小二乘支持向量机的造纸工控网络高隐蔽性入侵检测方法。首先,使用CEEMD算法对网络数据进行分解,得到一系列固有模态分量(IMF),利用排列熵对IMF分量进行分析,确定高噪声IMF分量;使用小波降噪对高噪声IMF分量展开抗干扰处理。然后,使用互信息特征选择方法对抗干扰处理后的入侵数据进行特征提取。最后,将提取到的入侵数据特征作为输入数据,通过最小二乘支持向量机(LS-SVM)建立一个判别函数,该函数根据输入数据的特征值进行分类,并判断网络中是否存在高隐蔽性入侵行为。实验结果表明,所提方法最高入侵检测准确率达到0.98,Kappa统计量最大为0.99,表明所提方法的数据处理效果好、网络入侵检测精度高。

Selective image enhancement method for low energy target information area

A selective subband enhancement method based on biorthogonal wavelet base is proposed. This novel image enhancement method is just for those images in which the energy of target information area is relatively lower. It includes two parts: one is enhancing the low frequency subband by wavelet decomposition and the other is building a new criterion based on entropy window to image evaluation. Experimental results show that this new scheme may result in a perfect image processing.

Fault Line Selection Method Considering Grounding Fault Angle for Distribution Network

In the distribution network system with its neutral point grounding via arc suppression coil, when single-phase grounding fault occurred near zero-crossing point of the phase voltage, the inaccuracy of the line selection always existed in existing methods. According to the characteristics that transient current was different between the fault feeder and other faultless feeders, wavelet transformation was performed on data of the transient current within a power frequency cycle after the fault occurred. Based on different fault angles, wavelet energy in corresponding frequency band was chosen to compare. The result was that wavelet energy in fault feeder was the largest of all, and it was larger than sum of those in other faultless feeders, when the bus broke down, the disparity between each wavelet energy was not significant. Fault line could be selected out by the criterion above. The results of MATLAB/simulink simulation experiment indicated that this method had anti-interference capacity and was feasible.

A data-mining approach to biomarker identification from protein profiles using discrete stationary wavelet transform

Objective: To develop a new bioinformatic tool based on a data-mining approach for extraction of the most infor- mative proteins that could be used to find the potential biomarkers for the detection of cancer. Methods: Two independent datasets from serum samples of 253 ovarian cancer and 167 breast cancer patients were used. The samples were examined by surface- enhanced laser desorption/ionization time-of-flight mass spectrometry (SELDI-TOF MS). The datasets were used to extract the informative proteins using a data-mining method in the discrete stationary wavelet transform domain. As a dimensionality re- duction procedure, the hard thresholding method was applied to reduce the number of wavelet coefficients. Also, a distance measure was used to select the most discriminative coefficients. To find the potential biomarkers using the selected wavelet coefficients, we applied the inverse discrete stationary wavelet transform combined with a two-sided t-test. Results: From the ovarian cancer dataset, a set of five proteins were detected as potential biomarkers that could be used to identify the cancer patients from the healthy cases with accuracy, sensitivity, and specificity of 100%. Also, from the breast cancer dataset, a set of eight proteins were found as the potential biomarkers that could separate the healthy cases from the cancer patients with accuracy of 98.26%, sensitivity of 100%, and specificity of 95.6%. Conclusion: The results have shown that the new bioinformatic tool can be used in combination with the high-throughput proteomic data such as SELDI-TOF MS to find the potential biomarkers with high discriminative power.

Selectivity estimation using compressed spatial information

Spatial selectivity estimation is one of the essential studies to get query responses rapidly and accurately with the limitation of memory space. Currently, there exist several spatial selectivity estimation techniques such as random sampling, histogram, and parametric. Especially, Cumulative Density Histogram guarantees accurate estimation for rectangle object which has multiple count problem. However, it requires large memory space because of retaining four sub histograms for spatial data. Therefore in this paper, we propose a new technique Cumulative Density Wavelet Histogram, called CDWH, which is the combination of Cumulative Density Histogram and Haar Wavelet Transform, a compressed technique. The proposed method simultaneously takes full advantage of their strong points, high accuracy provided by the former and economization of memory space supported by the latter. Consequently, our technique is able to support estimates with relatively low error and retain similar estimates even if memory space is small.

基于拉普拉斯小波滤波和SA-DS-CNN的滚动轴承故障诊断

针对基于深度学习模型的滚动轴承故障诊断方法易受环境噪声干扰的问题,提出了一种基于拉普拉斯小波滤波(LWF)和自注意力机制-动态选择-卷积神经网络(SA-DS-CNN)的滚动轴承故障诊断模型。首先,提出一种拉普拉斯小波阻尼参数自适应选取策略,使用拉普拉斯小波对采集的滚动轴承振动信号进行相关滤波并进行功率谱变换;其次,基于卷积神经网络框架,引入自注意力机制和动态选择机制,构造SA-DS-CNN;最后,利用SA-DS-CNN提取功率谱特征,根据轴承的不同故障状态定位相关特征信息,实现故障特征的提取和诊断。对N205EM圆柱滚子轴承的故障诊断试验表明:LWF降噪效果较好,能为SA-DS-CNN模型提供优秀的训练样本;SA-DS-CNN模型能抑制无用通道信息,增强网络特征学习能力;LWF和SA-DS-CNN组合模型的故障诊断准确率达到99.65%,优于其他组合模型。

数据驱动下农用车辆柴油机NO_(X)排放预测模型

针对农用车辆柴油机NO_(X)排放与实际运行工况之间的复杂非线性关系,提出一种数据驱动下的NO_(X)排放预测模型。基于车辆OBD采集实际运行数据,通过小波阈值降噪降低原始数据的非平稳性,采用集成特征选择算法完成模型输入特征的选择,同时融合BiGRU和注意力机制构成BiGRU-Attention模型,同时利用贝叶斯优化进行模型超参数选择。基于实车道路测试数据集分析,提出的模型相对于LSTM、GRU和BiLSTM-Attention模型NO_(X)瞬时排放预测校正系数分别提高7.65%、3.26%和4.09%,模型平均绝对误差维持在0.0014 g/s,在不同车辆数据集上预测校正系数均保持在85%以上,可以有效进行实际场景下NO_(X)排放的高精度预测,为农用车辆柴油机NO_(X)排放预测控制提供数据支撑。

基于时频域特征分析和ML-NN的故障电弧检测与选线

针对低压配电系统方式复杂、负载种类繁多、串联故障电弧的检测难度越来越大的问题,提出了1种基于时频域特征分析和多标签神经网络(ML-NN)分类的串联故障电弧检测与选线的方法.该方法通过采集多回路负载的不同支路发生电弧时的干路电流,对其时域采取统计的方法对故障电流的方差、均值、偏度和峰度进行分析,对其频域采用小波变换的方法得到其故障电流的小波系数特征.将时频域特征作为神经网络的输入进行训练,同时采用反向传播方法来训练模型,实现故障电弧检测和故障选线.经过实验验证,故障电弧检测和选线的准确度分别达到了97.57%、99%.

基于蚁群参数优化的LightGBM辐射源个体识别

为提升辐射源个体识别正确率和运算效率,提出了一种基于蚁群参数优化的LightGBM辐射源个体识别方法。运用提升小波包变换对辐射源信号数据进行特征提取并构建特征参数体系,对得到的特征数据集进行Z-score标准化处理;以最大分类正确率和最小特征子集规模为目标,建立了LightGBM参数优化和特征选择的数学模型;采用蚁群算法优化LightGBM的6个参数(最小叶子节点数据量、决策树的数量、学习率、L_(1)正则化项的权重、L_(2)正则化项的权重和最小叶子节点样本权重和);根据优化后的LightGBM得到每个特征的重要性值并使用序列后向搜索策略进行特征选择;最后通过LightGBM分类器实现对辐射源信号的分类识别。实验结果表明,所提方法在无噪声、信噪比为10 dB和信噪比为5 dB信号的数据集上的识别正确率优于对比特征选择方法GBDT、XGBoost和LightGBM的,同时特征维数的减少也提升了辐射源个体识别的运算效率。

基于GA优化BP神经网络的有源配电网高阻接地故障选线方法

当配电网发生高阻接地故障时,逆变型分布式电源的接入会向零序网络中注入不平衡的谐波电流,改变原有故障特征的分布规律,导致传统高阻故障选线方法失效。考虑光伏电源接入对配电网的影响,提出了一种基于GA优化BP神经网络通过融合多种故障特征的有源配电网高阻接地故障选线方法。首先,利用Matlab/Simulink搭建谐振接地系统仿真得到选定周波的故障零序电流,根据小波包变换从中提取小波包能量熵和模极大值,并将其作为数据样本。然后,将数据输入优化后的网络中进行训练,得到能够实现智能选线的机器学习模型。最后,算例分析表明该方法较传统算法提高了迭代速度和训练精度,在多种复杂故障条件下具有良好的选线容错率,且具有一定的抗噪能力。

基于电网谐波阻抗估计的并联电容器组优化配置方案

采用先投入高串抗率电容器、后投入低串抗率电容器这种单一配置方案,导致变电站出现高串率电容器频繁投切、寿命降低和无功容量浪费问题。为了解决这些问题,提出一种基于阻抗估计和多电能质量约束的变电站电容器组优化配置方法。首先,获取电容器投切后公共连接点母线电压电流数据;其次采用小波最大最小阈值去噪,通过Prony算法得到暂态电压、电流数据来估计系统谐波阻抗,最后结合AVC下达的调度指令和目标函数在满足电网电能质量约束条件下对电容器组进行优化配置。结果表明,在满足电能质量约束和AVC调度指令下,优化配置后高低串抗率配置频次均有所改善,高串抗率电容器组投入频次下降32%,低串抗率电容器组投入频次由26%提高至58%。由系统谐波数据、谐波源和电容器等元件组成的谐波阻抗模型,以及基于谐波谐振机理设计的电容器组配置方案,能够提高数据分析的准确性,减小电容器损坏以及解决无功容量损失问题,提高供电质量,可为变电站电容器组的配置提供借鉴。

基于可调Q因子小波变换的海杂波抑制算法

针对海杂波背景下弱目标检测中存在的信杂比低的问题,提出了改进的基于可调Q因子小波变换的海杂波抑制算法。由于海杂波能量远大于目标信号能量,提出选取与海杂波振荡特性相匹配的参数进行可调Q因子小波变换,得到各小波子带的系数,并对小波系数进行稀疏优化后重构海杂波信号。为了判断弱目标信号是否存在,提出一种自适应的阈值检测方法,将原始回波信号与海杂波重构信号的差作为检测样本,实现对弱目标信号的检测。该算法不依赖海杂波具体模型。最后对某实测海杂波数据集进行实验,验证了所提算法的正确性。