【作者】 张海龙；

【导师】 关根志；

【作者基本信息】 武汉大学 ， 高电压与绝缘技术， 2009， 博士

【摘要】 110～220kV XLPE电缆已成为电力系统的重要组成部分,XLPE电缆具有优良的介电性能和耐热性能、传输容量大、重量轻、电缆终端头结构简单、安装敷设方便。因此,随着电网的不断发展和城市电网改造的需要,XLPE电缆线路在电网中所占的比重日益增加。但是当其发生故障时,也会造成巨大的经济损失和严重的社会影响,所以,110～220kV XLPE电缆的安全运行对电力系统至关重要。随着状态检修观念的深入人心,在线检测技术得到了快速地发展。然而,XLPE电缆绝缘在线检测技术是电气设备绝缘在线检测领域长期存在的难题。特别是现有的低压XLPE电缆绝缘在线检测方法并不完全适用于110-220kV高压XLPE电缆。因此,本文针对110～220kV XLPE电缆的绝缘在线检测技术进行了系统的研究：1、对110-220kV高压XLPE电缆绝缘在线检测技术的研究意义、重要性及其发展历史进行综述,对XLPE电缆绝缘在线检测研究现状及检测方法进行评述,指出该领域研究的关键问题和发展方向。2、深入研究了XLPE电缆绝缘老化的基本理论。首先对水树枝的老化机理进行了研究。由研究可知水树枝长度随外加电压幅值和频率的增加而增加,针尖曲率半径越小,水树枝的长度越长；电缆的绝缘电阻、tanδ和接地线电流可以反应电缆的水树枝老化程度。然后对电树枝的老化机理进行研究,应用陷阱理论解释了XLPE电缆中电树枝的形成过程。最后对XLPE电缆中局部放电老化进行了理论研究。3、应用接地线电流法重点研究金属屏蔽层单端接地XLPE电缆绝缘在线检测方法。仿真研究表明电缆接地线电流和电缆的tanδ可以反映电缆的局部绝缘缺陷和电缆整体老化现象。提出以绝缘参数的发展趋势判断绝缘特性的诊断方法。并且应用虚拟仪器技术成功研制了基于接地线电流法的绝缘在线检测系统,该系统可以实时检测电缆的接地线电流、tanδ、绝缘等效电阻和绝缘等效电容。经过近三年的现场运行表明该系统稳定可靠,测量精确度高,方便运行人员及时了解电缆的绝缘状况。4、提出了一种新的金属屏蔽层交叉互联XLPE电缆绝缘在线检测方法。由于电缆的金属屏蔽层交叉互联,无法通过直接测量得到每根电缆的泄漏电流和tanδ,也就很难准确评判电缆的绝缘状况。因此本文提出通过测量A、B、C三相电缆首末端接地线电流的幅值和相位与电缆运行初期的接地线电流的幅值和相位之差反映每段电缆tanδ变化情况,以此反映电缆绝缘老化情况。利用向量图的分析方法论证了新方法的物理意义,根据电路理论推导出了电缆绝缘故障段位置与接地线电流幅值和相位变化量的对应关系。根据以上新方法建立了基于ART2A-E神经网络的交叉互联XLPE电缆绝缘故障段诊断算法,为交叉互联电缆的绝缘在线诊断开辟了新的途径。5、提出了基于宽带高频传感器和盲源分离算法的XLPE电缆局部放电信号提取技术。首先使用两个高频传感器分别测量流经试品和耦合电容的电流信号,然后应用改进盲源分离算法实现外界干扰信号与试品局部放电信号的分离。仿真试验和实验室模拟实验表明,该方法可以从局部放电和外界干扰信号的混合信号中准确的提取出局部放电信号。6、建立了以局部放电脉冲的时域负熵为X轴,以局部放电脉冲的频域负熵为Y轴的局部放电时频负熵谱图。计算机仿真实验和实验室模拟试验表明,不同局部放电类型的脉冲在时频负熵谱图中形成不同的聚类。由此,利用时频负熵谱图可以完成局部放电脉冲的分类,然后应用PCA算法实现局部放电脉冲的特征波形提取,并应用时频负熵谱图进行了局部放电脉冲的模式识别研究。更多还原

【Abstract】 110～220kV XLPE power cable has been an important component of the power system. XLPE power cable has many advantages, such as excellent dielectric properties and heat resistance, large transmission capacity, light weight, simple structure of cable terminations, and facilitated installation. With the development of power grid and the needs of urban power network renovation, the proportion of XLPE cable lines is increasing in power system. But its failure would cause huge economic losses and serious social impact so the safe operation of power systems is important for 110～220kV XLPE cable.As the concept of condition-based maintenance has been accepted by people, on-line monitoring technology has been developing rapidly. However, XLPE cable insulation on-line detection technology is always a difficult problem of insulation monitoring technique. Particularly, the existing low-voltage XLPE cable insulation on-line detection method cannot always adapt to the 110～220kV high-voltage XLPE cables. Therefore, it is necessary for 110～220kV XLPE cable to carry out the specialized research.1、It is a summary of 110～220kV high voltage XLPE cable insulation online detection that whose importance, research significance and the history development have been made in this paper. And the author has made comments on various online insulation detection methods and research status, in which the key problem and development has been pointed out.2、It is a basic theory which has been made a deep research into the XLPE cable insulation. The research shows that the water tree length is increasing as voltage amplitude and frequency, and the smaller the tip radius of curvature, the longer the length of water tree. The cable insulation resistance, tan 8 and Grounding Current can respond to the degree of water tree aging. Secondly, the aging mechanism of electrical trees has been studied and the trap theory has been applied to explain the XLPE power cables in the formation of branches. Finally, partial discharge of XLPE cables to the theoretical study of aging has been studied.3、It is Single-ended earthing of metal shield layer XLPE cable insulation on-line detection which is made focus research on with Grounding Current. The study of simulation results shows that Grounding Current and tanδof cable can reflect the cable insulation defects. Then, the method on diagnosing of insulation characteristics has been proposed. Finally, an insulation on-line detection system is successfully developed by the application of the virtual instrument technology and Grounding Current method. This system can be real-time detection of the grounding line current, tanδ, insulation equivalent resistance and insulation equivalent capacitance. Through three years’site operating, the system is stable, reliable and accurate. The operating personnel can realize the situation of cable insulation with convenience.4、A new insulation on-line monitoring method on metal shielding layer of cross--bonding XLPE cable has been proposed in this paper. Because the metal shield layers of the cable is crossing, the leakage current and tanδof a single cable cannot be obtained through direct measuring. So it is impossible to judge accurately the situation of the cable insulation. It is calculating amplitude difference and phase angle difference between XLPE cable grounding line current and initial installation of the current grounding line that through which the result reflects the changes in each cable insulation condition. By analyzing the vector diagram, the author demonstrates the Physical Meaning of the new method. According to the theory of circuit, it has been deduced the relationship between the cable insulation fault location with variables of grounding--line current amplitude and phase. Based on ART2A-E neural network, Cross-Bonding XLPE cable insulation fault section diagnosis algorithm has been created. It is a new way for Cross-Bonding XLPE Power cable insulation diagnosis.5、It has been put forward to the new partial discharge signal extraction technology of XLPE cable which is based on broadband high-frequency sensors and blind source separation algorithm. Firstly, current signal of test object and coupling capacitor have been measured respectively with two high-frequency sensors. And then it improves blind source separation algorithm to realize partial discharge signals’separation from outside interference signals. Simulation tests and laboratory simulation experiments show that this method can extracting accurate partial discharge signal from mixed-signal of outside interference and partial discharge signal.6. It is the first time that the negative entropy as the measurement of Partial Discharge Pulse Shape of sparse or non-Gaussian according to their characteristics. A new vector graph is built, in which time domain Negentropy of PD pulse is the X-axis and frequency domain Negentropy of PD pulse is the Y-axis. Simulation tests and laboratory simulation experiments show that different types of partial discharge pulse can form clusters in the time-frequency spectrum. So classification of partial discharge pulse can be completed through the spectrum. And then the application of PCA algorithm can realize the extraction of partial discharge pulse characteristics. All of that have laid the foundation of the partial discharge pulse pattern recognition.更多还原