基于小波奇异值和支持向量机的高压线路故障诊断

提出了基于小波奇异值（WSV）和支持向量机（SVM）的电力系统故障类型识别的新方法。利用WSV来量化故障特征,再与SVM结合进行故障类型识别。对故障线路三相电流信号进行小波包变换分解,获取故障信号的小波细节系数;利用相重构技术将小波细节系数向量形成系数矩阵,并对该矩阵作奇异值分解,获取小波奇异值;将小波奇异值向量输入到SVM分类器进行故障类型识别。仿真表明,对于不同的故障类型,其小波奇异值分布明显不同,而对于同一类型故障,其小波奇异值分布在不同的故障位置、过渡电阻的情况下仍保持很大的相似性。SVM具有训练样本少、训练时间短、识别率高等优点。     

基于小波奇异值和支持向量机的高压线路故障诊断

提出了基于小波奇异值(WSV)和支持向量机(SVM)的电力系统故障类型识别的新方法.利用WSV来量化故障特征,再与SVM结合进行故障类型识别.对故障线路三相电流信号进行小波包变换分解,获取故障信号的小波细节系数;利用相重构技术将小波细节系数向量形成系数矩阵,并对该矩阵作奇异值分解,获取小波奇异值;将小波奇异值向量输入到SVM分类器进行故障类型识别.仿真表明,对于不同的故障类型,其小波奇异值分布明显不同,而对于同一类型故障,其小波奇异值分布在不同的故障位置、过渡电阻的情况下仍保持很大的相似性.SVM具有训练样本少、训练时间短、识别率高等优点.     

In-situ diagnostics of HV outdoor insulators using laser-induced fluorescence spectroscopy

A new method for remote and non-intrusive diagnostics and surveillance of high-voltage insulators and insulation materials is presented, in particular pertinent to outdoor insulation. The method is based on laser-induced fluorescence (LIF) spectroscopy. The relationship between the LIF-spectrum and different materials and surface characteristics of silicone rubber samples are studied. These characteristics include hydrophobicity, filler content and biological growth. Furthermore, real insulators that have been naturally aged have also been studied. We conclude that laser-induced fluorescence has a significant potential to become a highly useful technique for contact-free and non-intrusive monitoring of the status of insulator materials in high-voltage systems.     

模拟电路故障诊断的邻近支持向量机集成方法

提出了基于非对称AdaBoost算法与邻近支持向量机的模拟电路集成故障诊断方法。该方法采用非对称AdaBoost解决邻近支持向量机处理多分类问题出现的数据不平衡问题,提高邻近支持向量机的分类准确率;利用高斯变异策略自适应选择核函数及正则化因子,提高集成邻近支持向量机的个体间差异性;最后获得分类精度高和泛化性能好的分类器。通过对模拟电路的故障诊断实验,验证了该方法的鲁棒性和可行性。诊断结果表明本文方法具有故障定位准确和泛化性好的特点。     

传感器非线性校正的遗传支持向量机方法

针对传感器非线性校正中现有的较为常用的神经网络法的不足和支持向量机参数难确定的问题,提出了一种遗传算法和支持向量机相结合的方法,阐述了支持向量机的非线性校正原理和遗传算法优化支持向量机参数的实现过程,并分别采用BP神经网络法和遗传支持向量机方法对压力传感器进行非线性校正.实验结果表明: BP神经网络法使得传感器的最大相...     

A new class of dual support vector machine NPID controller used for predictive control

A nonlinear proportional‐integral‐differential (NPID) controller used for predictive control tunes the proportional, integral, and differential gain coefficients according to the system prediction output. The prediction error has great influence on its performance. In this paper, we propose a dual support vector machine (SVM) NPID controller using an SVM prediction system output with less error. System model reflecting the system feature is constructed by SVM, based on the training dataset gained by system transfer function. The error compensation model is also constructed by SVM, based on prediction error with the above system model. The system prediction output is obtained by the aforementioned SVM models. And it is used to compute the PID gain coefficients. Simulation results of typical systems show that the proposed method has very little prediction error and high control performance. © 2015 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.     

基于支持向量机的随钻测量

针对目前随钻测量的需要,提出了一种基于支持向量机预测的随钻测量方法.与传统随钻测量方法相比该方法能够预测提示井下钻头需要待钻进的地理方位位置,为司钻人员下一步的施工钻进提供方位信息.通过现场试验数据测试,结果表明,该方法能够预测提示井下定向钻进过程中的定向方位信息,大幅度节省钻进施工时间.     

Optimization of HV electrode systems by neural networks using a newlearning method

To avoid a large number of iterations, optimization of electrode shapes has been done by artificial neural networks (NN). Two practical examples have been considered, an axisymmetric single-phase GIS bus termination and an axisymmetric transformer shield ring. The shape of the electrodes has been taken as quarter-ellipse or half-ellipse because an ellipse has more flexibility than a circle. For NN, the so-called resilient propagation algorithm, learning faster than the standard back-propagation algorithm, has been employed. The training sets as well as the test sets of NN have been prepared by charge simulation method     

采用模糊支持向量机的模拟电路故障诊断新方法

为了解决模拟电路故障诊断复杂多样难于辨识的问题,有效提高分类的准确度,提出了一种模拟电路故障诊断的新方法。首先对采集的信号进行时-频域联合特征提取并采用新的模糊隶属度函数确定训练样本的隶属度,消除噪声和野点对故障诊断的影响;然后将训练集输入模糊支持向量机分类方法训练获得故障诊断模型;最后将测试集输入训练好的模糊支持向量机分类模型,实现对不同故障类型的识别。将该方法应用于CTSV滤波电路进行故障诊断仿真实验,结果显示该方法不仅能正确分类单故障而且能有效分类多故障,平均故障诊断率达到98.2%,为模拟电路故障诊断开辟新的途径。     

基于支持向量机的电力系统状态估计 多类型数据异常检测

为了解决异常数据严重影响电力系统状态估计性能的问题,提出了一种基于支持向量机(SVM) 的电力系统预测辅助 状态估计(FASE) 多类型数据异常检测方法。首先,针对传统 FASE 的预测准确率欠佳的问题,提出了基于极限学习机的 FASE 方法,并利用SVM 并基于预测数据、量测数据与估计值,实现了对坏数据、负荷突变和单相接地等多种类型的数据异常 检测。其次,针对惩罚因子和核函数参数会影响分类精度的问题,提出采用灰狼算法对 SVM 参数进行优化,在兼顾计算速度 的同时提高了数据异常检测的准确率。最后,在IEEE33 和丹麦DTU7K47 节点主动配电网系统上进行仿真测试,所提方法 在正常工况下提升26.08%与26.76%,计算速度提升46.05%,在数据异常情况下准确率综合提升32.04%与29.27%,结果 表明,所提方法具备较强的通用性与实时性,可以有效地检测电力系统中各种类型的数据异常,并提高状态估计的性能。     

一种基于多类支持向量机的故障诊断算法

主要研究了多类支持向量机算法,及其在多类故障诊断问题中的运用.考虑到传统"一对一"算法和"一对多"算法的局限,提出了基于遗传算法的决策树支持向量机,利用遗传算法的全局随机搜索性能来构造决策树.根据样本数据的分布,遗传算法的适应度函数定义为两个子类集聚类中心之间的距离,使得在决策树的每一个节点最可分类别尽可能分开.通过对三个数据集进行仿真分析,表明该算法的性能优于两个传统算法,具有更高的推广能力,验证了算法的有效性.     

支持向量机在入侵检测系统中的应用

目前的入侵检测系统(IDS)存在着在先验知识较少的情况下推广能力差的问题。简述了IDS的基本原理,从本质上讲,入侵检测实际上是一个分类问题．就是通过检测把正常数据和异常数据分开。给出了入侵检测模型,论述了支持向量机(SVM)是在小样本学习的基础上发展起来的分类器设计方法．专门用于小样本数据．而且对数据维数不敏感．提出了基于SVM的通用入侵检测系统模型,它主要由审计数据预处理器、支持向量机分类器和决策系统3部分组成。说明了SVM系统模型的可行性、模型、工作过程、实现4方面的内容．     

Fault classification in SEIG system using Hilbert-Huang transform and least square support vector machine

This paper presents the transient performance analysis of self excited induction generator (SEIG) during both balanced and unbalanced faults using stationary frame d–q axis. Significance of fault detection and fault classification is also investigated in this study. Current signal of SEIG is extracted. Non stationary distorted current waveforms of SEIG during fault condition are considered as superimposition of various oscillating modes. To separate out these oscillating components known as intrinsic mode functions (IMFs), empirical-mode decomposition (EMD) is used. Hilbert transform (HT) is applied on the first four IMFs to extract instantaneous amplitude and frequency. Combination of EMD and HT is known as Hilbert-Huang transform. To classify different faults of SEIG system, least square support vector machine (LSSVM) is used. Finally the superiority of the proposed SVM is established through comparison with support vector machine and probabilistic neural network.     

Flashover of discontinuous pollution layer on HV insulators

This paper deals with discontinuous pollution layers deposited on the insulator surface. Two zones are known to form: a clean (dry) zone and a polluted (wetted) zone. The influence of the length and position of the dry zone on the leakage current and the arc length are investigated. These were made on a plane experimental model under ac voltage. Two configurations were studied: pollution deposited near the HV electrode, and pollution deposited near the ground electrode. This study was made using pollution layers with different conductivities. Based on the experimental results, a theoretical model allowing us to calculate the gap impedance, the transferred (gap) voltage and the leakage current was established. This model enables to predict the pollution severity based on the sample's equivalent impedance     

基于Hilbert-Huang变换和最小二乘支持向量机的电力市场出清电价预测

随着世界各国电力工业改革的发展趋势,我国于20世纪90年代也开始了以打破垄断、引入竞争、放松管制为目标的电力市场化改革。如何合理制定相应的运营模式以及怎样根据电力市场的相关历史数据准确的预测出未来的市场出清电价,对于市场中的各个参与者都具有十分重要的意义。而实际电力市场的出清电价数据具有很强的非平稳性,Hilbert-Huang变换是分析处理非平稳性信号数据非常有效的方法,本文应用Hilbert-Huang变换首先对电力市场出清电价数据进行平稳化处理,然后运用最小二乘支持向量机（LS-SVM）对处理后的数据进行预测。预测结果表明,此模型显著的提高了出清电价预测的精度。     

基于支持向量回归机的谐波分析

传统谐波分析方法中,窄带滤波器选频法对元件参数十分敏感,受外界环境影响较大,且精度难以保证;快速傅里叶分解法会产生频谱泄漏误差,不同频率谐波之间也有干扰;神经网络分析方法检测精度高,但其基于大样本的学习方法,受样本及训练初始值影响较大,有时会导致训练不成功。为减小外界环境影响,提高检测精度和降低噪声影响,提出基于支持向量回归机的谐波分析方法,它是基于统计学习理论,以结构风险最小化为原则的机器学习,通过引入松弛变量和损失函数提高算法泛化能力和减小误差,该算法最终转化为标准二次规划问题,有全局最优解。通过算例分析,并与傅里叶检测方法相比较,可知该算法具有稳定性好、检测精度高、对噪声不敏感等优点。     

Optimal locality preserving least square support vector machine

In this paper, a novel least square support vector machine (LSSVM), termed as optimal locality preserving LSSVM (OLP-LSSVM) is proposed. By integrating structural risk minimization and locality preserving criterion in a unified framework, the resulting separating hyperplane is not only in accordance with the structural risk minimization principle but also be sensitive to the manifold structure of data points. The proposed model can be solved efficiently by alternating optimization method. Experimental results on several public available benchmark datasets show the feasibility and effectiveness of the proposed method.     

基于支持向量机模型的环境音分类研究

音频分类是提取音频结构和内容语义的重要手段,是基于内容的音频、视频检索和分析的基础.支持向量机(SVM)是一种有效的统计学习方法.本文提出了一种基于SVM的音频分类算法.将环境音分为6类:车鸣声,钟声,风声,冰块声,机床声和雨声.特征抽取是音频分类的基础.本文从帧层次上深入分析了不同类音频之间的区别性特征,包括频域能量,子带能量,过零率,频率中心,带宽,基音频率及MFCC(Mel-Frequency Cepstral Coefficients).实验结果表明,支持向量机模型的环境音分类性能较好,最优分类精度达到97.73%.     

A new flashover dynamic model of polluted HV insulators

In this work, a model based on field criterion has been developed to represent the flashover phenomenon, which occurs due to surface pollution on high voltage insulators, under ac voltage. The values of potential and electric field on an insulator surface have been determined using the finite element method (FEM). The open model of the insulator has been used for calculating the resistance in series with the arc, in addition to the values of the leakage current and the arc gradient. As a new approach, this dynamic model uses Lagrange multipliers for the solution of the pollution flashover problem. Both the impedance and the electric field criterion have been used for the propagation of arc on the surface. A computer program called NFDM (new flashover dynamic model) has been developed to achieve this. The results obtained from the program have been compared with theoretical and experimental results of other researchers.     

One airport detection method based on support vector machine

This paper proposes a novel airport detection method, which integrates the texture features and shape features of the airport. Eight texture features, such as the mean of the region, the deviation of the region, the smoothness of the region, the skewness of a histogram, the uniformity of the region, the randomness of the region, the mean of the gradient image and the deviation of the gradient image, are used to represent the features of the region. In this method, first the long lines are detected and the regions where the lines locate are segmented. Second, support vector machine (SVM) based on Gaussian kernel is used as a classifier which discriminates the runway from other candidate regions. Experimental results show that the error rate of the proposed method is lower than those of conventional methods which detect airport only by the shape feature of runway. The detection accuracy of the proposed method is nearly ten times higher than that of Liu’s methods, and the method has favorable speed for a real-time system. __________ Translated from Journal of Xi’an Jiaotong University, 2006, 40(6): 709–713 [译自: 西安交通大学学报]