Hilbert-Huang变换在电力系统过电压识别中的应用

将希尔伯特-黄变换(HHT)方法引入到过电压信号分类识别研究中,该方法由经验模态分解(EMD)与Hilbert变换两部分组成,能够定量、准确地对信号幅值及频率特征进行刻画。采用HHT方法对7种电力系统中常见的过电压信号进行了分析计算,结果表明HHT方法计算得到的瞬时幅值谱、Hilbert边际谱、Hilbert时频谱,能够作为特征量对不同种类型过电压进行分类识别。基于HHT算法与RBF神经网络搭建了过电压信号分层识别系统,并以实测过电压数据对其进行验证,结果表明,该分类识别系统能够有效地对过电压信号进行分类识别,并具有较高的识别率。     

基于改进变分模态分解和Hilbert变换的变压器局部放电信号特征提取及分类

针对现有局部放电(PD)信号特征提取方法存在的不足,提出一种基于变分模态分解(VMD)和Hilbert变换(Hilbert-VMD)的特征提取方法,并提出一种双阈值筛选法来确定VMD算法中的分解模态数。首先,根据PD信号功率谱,采用双阈值筛选法确定VMD算法中的分解模态数;其次,采用VMD算法对PD信号进行分解,得到数个有限带宽的固有模态分量(BLIMFs);然后,对各模态分量进行Hilbert变换并线性叠加后得到PD信号的Hilbert时频谱,并计算各模态分量的边际谱;最后,根据各模态分量的边际谱提取PD信号频域内的特征量,并用支持向量机(SVM)对所提取的特征量进行分类。实验结果表明,对试验环境下和现场实测两种环境下的PD信号,采用该文方法提取得到的特征量均具有较高的正确识别率,充分说明该特征提取方法可以有效提取PD信号特征。对于噪声较大的实测信号,采用该方法得到的正确识别率并未明显降低,说明该方法具有较好的噪声鲁棒性。此外,该文所提Hilbert-VMD方法也为PD信号提供了一种新的时频分析方法。     

应用改进Hilbert-Huang变换下的Volterra模型诊断OLTC机械故障

为提高有载分接开关(OLTC)机械故障诊断的自适应性、特征分辨率以及识别效率,提出一种包含聚合经验模态分解(EEMD)分解和Hilbert边际谱分析的改进HHT方法,与混沌时间序列的Volterra模型相结合来提取OLTC的机械故障特征。具体应用时,首先对OLTC切换过程中的多通道振动信号进行EEMD分解得到固有模态函数(IMF)分量,然后应用Hilbert谱分析法求取各IMF的Hilbert边际谱。进一步,应用Volterra模型根据Hilbert边际谱构建Volterra特征矩阵,以矩阵奇异值为故障特征参量。最后搭建了OLTC典型机械故障真型实验平台,采用文中方法获取并分析了几种典型机械故障的振动信号,并借助多分类支持向量机对数据集进行分类识别,验证了所提出故障诊断方法的有效性。与其他方法对比得知,新方法取得了较高的故障识别准确率。     

基于粗糙集神经网络和振动信号的高压断路器机械故障诊断

为了准确检测出高压断路器的机械故障类型,该文提出一种基于本征模态边际谱能量与粗糙集神经网络相结合的高压断路器振动信号故障诊断方法。首先将断路器的振动信号经过经验模态分解(EMD),得到若干个本征模态函数(IMF),对各个IMF分量进行希尔伯特(Hilbert)变换得到Hilbert边际谱,求取Hilbert边际谱的二次方得到Hilbert边际谱能量作为特征向量。基于粗糙集理论对特征向量进行属性约简分析,从而建立简单明了的决策表,根据决策表规则建立径向基函数(RBF)神经网络故障模型。实验结果表明,该方法能有效对高压断路器的机械故障类型进行分类。     

基于VMD-WVD分布与堆栈稀疏自编码网络的局放类型识别

针对目前常用的浅层模式识别方法对高维大容量样本处理困难的问题,提出一种基于变分模态分解(variational mode decomposition,VMD)的Wigner Ville分布(Wigner-Ville distribution,WVD)和堆叠稀疏自编码网络(stockedsparse auto-encoder,SSAE)的局部放电(partial discharge,PD)信号的模式识别方法。首先,以VMD算法对PD信号进行分解,对所得各分量进行时频分析得到相应的WVD;然后,以PD信号的VMD-WVD分布为输入量,利用SSAE对样本集合进行训练,自主提取内在特征。此外,将SSAE与稀疏自编码器(stackedsparseauto-encoder,SAE)的输出特征进行比较,验证了SSAE网络特征提取能力的优越性;最后,用训练好的SSAE网络完成测试样本的局部放电类型的识别。同时,以基于反向传播(backpropagation,BP)神经网络和支持向量机(support vector machine,SVM)的识别结果与该结果进行比较。结果表明,所采用的识别方法具有更高的正确识别率。     

基于HHT的电能质量扰动定位与分类

针对电能质量扰动定位和识别分类的需求,提出了一种基于HHT的电能质量扰动定位与分类的新方法。采用HHT算法对电能质量扰动信号进行变换,获得瞬时幅值、Hilbert谱和边际谱,并利用Hilbert谱对扰动信号进行定位。从瞬时幅值、Hilbert谱和边际谱中提取特征量,为决策分类树提供判断依据以便进行分类识别。仿真实验结果表明,采用HHT算法与决策分类树相结合的电能质量扰动定位与分类不需训练,提取的特征量少而有效,分类识别的效果较好,具有良好的抗噪性能。     

基于稀疏自动编码器深度神经网络的电能质量扰动分类方法

针对智能电网日益突出的电能质量扰动问题,提出了一种基于稀疏自动编码器(SAE)深度神经网络的电能质量扰动分类方法。利用SAE对电能质量扰动原始数据进行无监督特征学习,自动提取数据特征的稀疏特征表达;通过堆栈式稀疏自动编码器(SSAE)进行逐层学习,获得电能质量扰动数据的深层次特征;将其连接到softmax分类器进行微调训练,并输出电能质量扰动事件分类结果。利用已添加高斯白噪声的数据对SSAE进行训练,以提高其特征表达的抗噪声能力。仿真结果表明,所提方法能够准确地识别包含2种复合扰动在内的9种电能质量扰动信号,并且具有很好的鲁棒性。     

基于EMD的Hilbert变换应用于暂态信号分析

将一种新的非平稳信号处理方法--基于经验模态分解(EMD)的希尔伯特(Hilbert)变换方法,应用于电力系统暂态信号分析中。通过EMD方法提取信号的固有模态函数(IMF),再进行Hilbert变换,求瞬时频率、瞬时振幅,得到信号的Hilbert谱,进而得到Hilbert边际谱,对故障暂态和扰动信号进行了分析。通过瞬时频率进行故障暂态和扰动时刻的准确检测;通过Hilbert边际谱与傅里叶幅值谱的比较,表明Hilbert边际谱在分辨率上具有明显的优越性。该方法为电力系统暂态信号分析提供了一种新的分析手段。仿真结果验证了该方法的有效性。     

基于EMD的Hilbert变换应用于暂态信号分析

将一种新的非平稳信号处理方法——基于经验模态分解(EMD)的希尔伯特（Hilbert）变换方法，应用于电力系统暂态信号分析中。通过EMD方法提取信号的固有模态函数（IMF），再进行Hilbert变换，求瞬时频率、瞬时振幅，得到信号的Hilbert谱，进而得到Hilbert边际谱，对故障暂态和扰动信号进行了分析。通过瞬时频率进行故障暂态和扰动时刻的准确检测；通过Hilbert边际谱与傅里叶幅值谱的比较，表明Hilbert边际谱在分辨率上具有明显的优越性。该方法为电力系统暂态信号分析提供了一种新的分析手段。仿真结果验证了该方法的有效性。     

基于Hilbert边际谱和IPSO-SVDD的滚动轴承故障诊断

滚动轴承是旋转机械状态监控及故障诊断的重要研究内容。为了更加高效的对轴承故障位置及故障程度进行诊断,提出了一种基于Hilbert边际谱和改进粒子群算法(IPSO)优化支持向量数据描述(SVDD)相结合的滚动轴承故障诊断方法。该方法首先求取轴承振动信号的本征模态函数,在此基础上得到信号的边际谱以及信号的AR模型参数,积分求取边际谱的能量特征函数和AR模型参数相结合构成系统特征向量。然后针对传统网格搜索法或凭经验确定SVDD核心参数的缺点,提出利用基于动态因子的粒子群算法对SVDD的核心参数惩罚常数C及核函数宽度σ进行优化,利用优化后的SVDD模型对滚动轴承各状态信号进行智能诊断。人工数据集及真实数据集实验结果表明,该方法可以有效识别各故障状态信号,并且优化后模型的诊断效率及诊断精度高于传统网格搜索法确定的模型。     

Fast recognition of bird sounds using extreme learning machines

Recognition of bird species by their sounds can bring considerable significance to both ecologists and ornithologists for measuring the biodiversity in the reserves, and studying climate changes. In this letter, we propose an efficient method based on an extreme learning machine (ELM) to classify bird sounds of 86 species of birds in very limited training and testing time. Experimental results prove that, the proposed ELM method can achieve the best recognition performance (81.1 %, unweighted average recall) compared with K‐nearest neighbours (K‐NN), support vector machines (SVM), neural networks (NN), and deep neural networks (DNN) pre‐trained by an autoencoder. In addition, ELM requires the least total time for training and testing (2.047 ± 0.034 s). © 2016 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.     

Discrimination between PD pulse shapes using different neuralnetwork paradigms

A comparison has been carried out on the partial discharge (PD) pulse shape recognition capabilities of neural networks, using the nearest neighbor classifier, learning vector quantization and multilayer perceptron paradigms. The PD pattern recognition capabilities were assessed on artificial cylindrical cavities of different sizes. The performance of the three neural network paradigms was found to be equivalent in all respects, with the exception of the case where a distinction was required between small cavity sizes; under those circumstances, the learning vector quantization paradigm was distinctly superior to the two other paradigms. The experimental results also demonstrated that, even with simple metallic electrode cavities, the discrimination capabilities of the three types of neural networks are not always perfect     

采用Hilbert-Huang变换和支持向量机的过电压分类识别方法(英文)

This paper proposes an effective method for over-voltage classification based on the Hilbert-Huang transform(HHT) method.Hilbert-Huang transform method is composed of empirical mode decomposition(EMD) and Hilbert transform.Nine kinds of common power system over-voltages are calculated and analyzed by HHT.Based on the instantaneous amplitude spectrum,Hilbert marginal spectrum and Hilbert time-frequency spectrum,three kinds of over-voltage characteristic quantities are obtained.A hierarchical classification system is built based on HHT and support vector machine(SVM).This classification system is tested by 106 field over-voltage signals,and the average classification rate is 94.3%.This research shows that HHT is an effective time-frequency analysis algorithms in the application of over-voltage classification and identification.     

深度学习辅助的区域交直流配电网区间状态估计

针对区域交直流混合配电网中实时量测覆盖率低、量测误差分布具有不确定性的问题,提出了基于深度神经网络(DNN)伪量测建模的交直流配电网区间状态估计方法。该方法首先对DNN进行离线训练,然后将实时量测数据和电压源换流器控制的变量值作为DNN的输入特征,建立伪量测模型;接着,在实时量测更新时,利用已训练好的DNN快速生成伪量测;最后,对伪量测和实时量测的不确定性采用区间形式建模并进行区间状态估计,进而准确监测交直流系统状态。算例仿真结果表明,所提方法能够避免对量测误差的概率分布进行假设,并且能够在低冗余量测配置或量测缺失时,准确获得交直流配电网状态变量的上下界信息。     

基于二维变分模态分解和Hilbert变换的局放信号特征提取方法

本文通过对二维Hilbert-Huang变换方法的改进,提出了一种基于二维变分模态分解(VMD)和Hilbert变换的局部放电灰度图像特征提取方法。首先,利用局部放电样本生成相应放电灰度图;其次,以二维VMD算法分解各放电灰度图像,获取各个不同中心频率的模态分量;然后,通过四元数Hilbert变换得到各模态函数对应的特征图,并提取灰度纹理特征,构成各放电样本对应的特征向量;最后,以BP神经网络分类器对提取出的局部放电特征量进行分类和识别。实验结果验证表明,同二维Hilbert-Huang变换和传统放电灰度图特征提取方法相比,基于本文方法所得特征量具有更高的正确识别率,验证了该方法的可行性。另外,本文所采用的二维VMD-Hilbert方法为局部放电信号的频谱分析拓展了新的思路。     

Signal Processing and Deep Learning Techniques for Power Quality Events Monitoring and Classification

Abstract

Power quality disturbances (PQDs) have major challenges in embedded generation systems, renewable energy networks, and HVDC/HVAC electrical power transmission networks. Due to PQDs, electrical power network can have disruption in the protection system, security system, and energy-saving system. PQDs also affect the operation cost and consistency of electrical power systems. This paper presents an innovative method based on compressive sensing (CS), singular spectrum analysis (SSA), wavelet transform (WT) and deep neural network (DNN) for monitoring and classification of PQDs. Feature extraction and selection is an essential part of the classification of PQDs. In this paper, initially, SSA time-series tool and multi-resolution wavelet transform are introduced to extract the features of PQDs, and then CS technique is used to reduce the dimensionality of the extracted features. Finally, DNN-based classifier is used to classify the single-and-combined PQDs. The DNN architecture is constructed utilizing the restricted Boltzmann machine, which is then fine-tuned by back-propagation. The heart of this paper is to enhance the classification and monitoring accuracy and comparison of the results of WT-based classifier with SSA-based classifier. The proposed method is tested using 15 types of single and combined PQDs. These disturbances are transitory in the transmission and distribution networks such as voltage sag, swell, transient, interruption, harmonic, etc. The simulation and experimental results demonstrate that the SSA-based DNN classifier has significantly higher potential than the WT-based classifier to classify the power quality events under noisy and noiseless conditions.     

Fuzzy learning vector quantization networks for power transformercondition assessment

To improve the assessment capability of power transformers, this paper proposes a new intelligent decision support system based on fuzzy learning vector quantization (HVQ) networks. In constructing the system, a fuzzy-based classifier is designed to divide the historical data for dissolved gas analysis (DGA) into various categories with different levels of gas attributes. For each category of gas attributes, a learning vector quantization (LVQ) network is trained to be responsible for the classification of the potential faults due to insulation deterioration. The assessment approach has been tested on the DGA data from Taiwan Power Company (TPC) and compared with the previous fuzzy diagnosis system and the existing multi-layered backpropagation based artificial neural networks (BPANN) methods. Remarkable classification accuracy and far less training efforts of the proposed approach are achieved in this paper