输电线路雷击仿真与识别方法研究
发布时间：2018-08-21 20:46
【摘要】：高压输电系统是电力系统中承担着电能传输任务的重要组成部分，雷击是造成输电线路绝缘事故的主要原因之一，对电力系统的危害极大。为了防止雷害故障的频繁发生，在现场要求采取的措施一定要有针对性，要针对不同的雷害故障，采取不同的措施。因此不论是进行故障处理还是电力系统的防雷改进，都必须首先分析清楚雷击的类型，，否则必然是劳而无功，雷害事故依旧频繁发生。因此有必要对雷击类型的识别进行研究。本文对雷击220kV输电线路进行了仿真分析，获取不同类型雷击的过电压波形，提取特征量，结合智能分类方法实现对雷击的分类识别。 采用PSCAD/EMTDC暂态仿真软件进行220kV输电线路雷击仿真。考虑了杆塔的分段波阻抗模型，对输电线路雷击进行了建模分析。得到普通短路故障、感应雷击、直击雷干扰、反击故障和绕击故障5种扰动下的仿真波形。对仿真得到的电压波形进行了初步的分析，并结合过电压产生机理，分析了不同类型雷击电压信号的时域波形差异。 通过时域、频域、时频联合分析方法对过电压信号进行分析，得到了全面反映信号特点的特征量。基于时域波形提取了三相波形相似度W、电压积分特征值ω和波头波动性B三个特征量；基于希尔伯特-黄变换和Hilbert时频谱提取了反映频域和时频关联的特征量Q1、Q2、Q3、Q4；基于teager能量算子提取了反映波头变化特性和陡度的特征量Tg和Tm。 考虑到不同扰动的特点和各个特征量的物理意义，为每类扰动的识别选择相应的特征量，以提高识别的准确率。雷击与普通短路故障的差异主要体现在波头波动性B和高低频能量比值Q2上，雷击的波头波动性大，高频分量含量高；感应雷击同一时刻出现在输电线路的三相导线中，所以三相波形相似度W和Hilbert能谱矩阵相似度Q3较高，可区分感应雷击和直击雷击。雷击干扰的电压信号附加分量具有周期正负交变性，故电压积分特征值ω可用于识别雷击干扰。绕击和反击的差异体现在波头特征和高频能量分布特征，通过计算反映波头陡度的Teager能量值，Teager能量值的最大值Tm及反映最大值是否出现在第一个极值点的特征量Tg，结合高频能量分布特征量Q4可有效的区分反击与绕击。 本文研究了相关向量机的基本概念、算法及核函数的选择。利用多个二分类相关向量机建立了输电线路雷击类型的识别模型。将提取的特征量输入分类器，为每个RVM选择相应的特征向量。采用仿真得到的数据对识别模型进行训练和测试。最终实现对输电线路雷击类型的智能识别。大量的仿真分析验证了所建识别模型的有效性。
[Abstract]:High voltage transmission system is an important part of power transmission task in power system. Lightning strike is one of the main causes of transmission line insulation accidents, which is very harmful to power system. In order to prevent the frequent occurrence of lightning damage faults, the measures required in the field must be targeted, and different measures should be taken according to different lightning damage faults. Therefore, it is necessary to first analyze the type of lightning strike for fault treatment or lightning protection improvement of power system, otherwise, it is inevitable that there is no work and lightning accidents still occur frequently. Therefore, it is necessary to study the recognition of lightning strike type. In this paper, the 220kV transmission lines are simulated and analyzed, the overvoltage waveforms of different types of lightning strikes are obtained, the characteristic quantities are extracted, and the classification and recognition of lightning strikes are realized with intelligent classification method. PSCAD/EMTDC transient simulation software is used to simulate lightning stroke of 220kV transmission line. Considering the sectional wave impedance model of the tower, the lightning stroke of transmission line is modeled and analyzed. The simulation waveforms of five kinds of disturbances, such as common short circuit fault, inductive lightning strike, direct strike lightning disturbance, counterattack fault and wound failure, are obtained. The voltage waveform obtained by simulation is analyzed preliminarily, and the time-domain waveform difference of different types of lightning voltage signals is analyzed in combination with the mechanism of over-voltage generation. The overvoltage signal is analyzed by time-domain, frequency-domain and time-frequency joint analysis method, and the characteristic quantity reflecting the characteristics of the signal is obtained. Based on the time domain waveform, three characteristic variables, namely, the similarity of three-phase waveform W, the voltage integral eigenvalue 蠅 and the wave head volatility B, are extracted, and the characteristic quantity Q _ 1 / Q _ 2Q _ 3Q _ 3 / Q _ 4, which reflects frequency domain and time-frequency correlation, is extracted based on Hilbert-Huang transform and Hilbert time-frequency spectrum. Based on the teager energy operator, the characteristic quantities TG and TM, which reflect the variation characteristics and steepness of the wave head, are extracted. Considering the characteristics of different disturbances and the physical meaning of each characteristic quantity, the corresponding feature variables are selected for the recognition of each kind of disturbance, so as to improve the accuracy of recognition. The difference between lightning stroke and common short-circuit fault is mainly reflected in wave head fluctuation B and ratio Q _ 2 of high and low frequency energy, the wave head of lightning stroke is highly volatile and the content of high frequency component is high, and the inductive lightning strike appears in the three-phase conductor of transmission line at the same time. Therefore, the similarity of three-phase waveform W and Hilbert energy spectrum matrix Q _ 3 is high, which can distinguish induced lightning stroke from direct lightning stroke. The additional components of the voltage signal of lightning disturbance have positive and negative period alternation, so the voltage integral eigenvalue 蠅 can be used to identify the lightning disturbance. The difference between wound strike and counterattack is reflected in the characteristics of wave head and high frequency energy distribution. By calculating the maximum value T _ m of Teager energy value and whether or not the maximum value appears in the first extreme point, the characteristic quantity Q4 of high frequency energy distribution can be used to distinguish the striking back from the round strike. The maximum value of Teager energy and the maximum value of Teager energy reflect the steepness of the wave head and whether the maximum energy value appears in the first extreme point is calculated. In this paper, the basic concept, algorithm and kernel function selection of correlation vector machine are studied. The identification model of lightning strike type of transmission line is established by using multiple binary correlation vector machines. The extracted features are input into the classifier and the corresponding feature vectors are selected for each RVM. The recognition model is trained and tested by the simulation data. Finally, the intelligent identification of lightning strike type of transmission line is realized. A large number of simulation results show the effectiveness of the proposed model.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM863
本文编号：2196305
[Abstract]:High voltage transmission system is an important part of power transmission task in power system. Lightning strike is one of the main causes of transmission line insulation accidents, which is very harmful to power system. In order to prevent the frequent occurrence of lightning damage faults, the measures required in the field must be targeted, and different measures should be taken according to different lightning damage faults. Therefore, it is necessary to first analyze the type of lightning strike for fault treatment or lightning protection improvement of power system, otherwise, it is inevitable that there is no work and lightning accidents still occur frequently. Therefore, it is necessary to study the recognition of lightning strike type. In this paper, the 220kV transmission lines are simulated and analyzed, the overvoltage waveforms of different types of lightning strikes are obtained, the characteristic quantities are extracted, and the classification and recognition of lightning strikes are realized with intelligent classification method. PSCAD/EMTDC transient simulation software is used to simulate lightning stroke of 220kV transmission line. Considering the sectional wave impedance model of the tower, the lightning stroke of transmission line is modeled and analyzed. The simulation waveforms of five kinds of disturbances, such as common short circuit fault, inductive lightning strike, direct strike lightning disturbance, counterattack fault and wound failure, are obtained. The voltage waveform obtained by simulation is analyzed preliminarily, and the time-domain waveform difference of different types of lightning voltage signals is analyzed in combination with the mechanism of over-voltage generation. The overvoltage signal is analyzed by time-domain, frequency-domain and time-frequency joint analysis method, and the characteristic quantity reflecting the characteristics of the signal is obtained. Based on the time domain waveform, three characteristic variables, namely, the similarity of three-phase waveform W, the voltage integral eigenvalue 蠅 and the wave head volatility B, are extracted, and the characteristic quantity Q _ 1 / Q _ 2Q _ 3Q _ 3 / Q _ 4, which reflects frequency domain and time-frequency correlation, is extracted based on Hilbert-Huang transform and Hilbert time-frequency spectrum. Based on the teager energy operator, the characteristic quantities TG and TM, which reflect the variation characteristics and steepness of the wave head, are extracted. Considering the characteristics of different disturbances and the physical meaning of each characteristic quantity, the corresponding feature variables are selected for the recognition of each kind of disturbance, so as to improve the accuracy of recognition. The difference between lightning stroke and common short-circuit fault is mainly reflected in wave head fluctuation B and ratio Q _ 2 of high and low frequency energy, the wave head of lightning stroke is highly volatile and the content of high frequency component is high, and the inductive lightning strike appears in the three-phase conductor of transmission line at the same time. Therefore, the similarity of three-phase waveform W and Hilbert energy spectrum matrix Q _ 3 is high, which can distinguish induced lightning stroke from direct lightning stroke. The additional components of the voltage signal of lightning disturbance have positive and negative period alternation, so the voltage integral eigenvalue 蠅 can be used to identify the lightning disturbance. The difference between wound strike and counterattack is reflected in the characteristics of wave head and high frequency energy distribution. By calculating the maximum value T _ m of Teager energy value and whether or not the maximum value appears in the first extreme point, the characteristic quantity Q4 of high frequency energy distribution can be used to distinguish the striking back from the round strike. The maximum value of Teager energy and the maximum value of Teager energy reflect the steepness of the wave head and whether the maximum energy value appears in the first extreme point is calculated. In this paper, the basic concept, algorithm and kernel function selection of correlation vector machine are studied. The identification model of lightning strike type of transmission line is established by using multiple binary correlation vector machines. The extracted features are input into the classifier and the corresponding feature vectors are selected for each RVM. The recognition model is trained and tested by the simulation data. Finally, the intelligent identification of lightning strike type of transmission line is realized. A large number of simulation results show the effectiveness of the proposed model.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM863
【参考文献】
相关期刊论文 前10条
1 王安定,葛耀中;模量变换技术在反应故障分量的微机保护中的应用研究[J];电力系统自动化;1988年03期
2 郭宁明;覃剑;陈祥训;;雷击对行波故障测距的影响及识别[J];电力系统自动化;2008年05期
3 段青;赵建国;马艳;;基于稀疏贝叶斯学习的电力系统暂态稳定评估[J];电力自动化设备;2009年09期
4 叶会生,何俊佳,李化,林福昌;雷击高压直流线路杆塔时的过电压和闪络仿真研究[J];电网技术;2005年21期
5 邹贵彬;高厚磊;王明军;朱峰;杨荣华;;一种雷电波侵入变电站的扰动识别方法[J];电网技术;2012年04期
6 丁美新,李慧峰,朱子述,肖登明;雷电流波形的数学模型及频谱仿真[J];高电压技术;2002年06期
7 吴昊;肖先勇;;基于小波变换和行波理论的输电线路雷击定位[J];高电压技术;2007年01期
8 吴昊;肖先勇;邓武军;;输电线行波测距中雷击与短路故障的识别[J];高电压技术;2007年06期
9 陈家宏;张勤;冯万兴;方玉河;;中国电网雷电定位系统与雷电监测网[J];高电压技术;2008年03期
10 肖先勇;李逢;邓武军;;雷击与短路故障的S变换特征量识别方法[J];高电压技术;2009年04期
本文编号：2196305
本文链接：https://www.wllwen.com/kejilunwen/dianlilw/2196305.html
