基于电场特征集和支持向量机的空气间隙起晕和击穿电压预测研究

发布时间：2018-05-12 07:28

本文选题：空气间隙 + 起晕电压　；参考：《武汉大学》2014年博士论文

【摘要】：特高压电网可以显著提高电网的安全性、可靠性、灵活性和经济性。电磁环境和外绝缘特性是特高压输变电工程设计和运行的重点关注问题,其核心内容分别为电晕控制和空气间距选择,基本依据分别为空气间隙的起晕和击穿特性。考核性试验是检验特高压工程设计是否满足要求的主要依据,但试验存在周期长、代价高的问题,同时难以穷举实际输变电工程间隙。此外,试验取得的研究结论大都建立在间隙尺度之上,对于工程上难以采用几何尺度进行描述的复杂间隙结构,其适用范围受到了很大的限制。因此,有必要深入开展空气间隙放电机理和模型的研究,通过仿真手段预测得到各种间隙结构在不同运行条件下的起晕和击穿电压值,并用于指导实际输变电工程设计。目前,空气间隙起晕和击穿电压的预测方法主要包括经验、半经验公式和物理模型。经验、半经验公式由于某些参数是在特定试验条件或前提假设下获得的,故适用范围有限。物理模型从放电发展的物理过程来预测电晕或击穿特性,在过去20年内取得了长足进展。然而,由于物理模型十分复杂,仍存在一些尚未解决的问题,导致其预测值与试验值仍存在较大偏差,难以有效指导实际工程设计。针对上述问题,本文提出了一种基于电场特征集和支持向量机(support vector machine, SVM)的空气间隙起晕和击穿电压预测方法。基于提出的方法,对输变电工程中的金具(如棒、球结构)和绞线的直流起晕电压以及不同电极结构空气间隙的击穿电压进行了预测,并分析了相关的影响因素。同时,对部分间隙结构的起晕和击穿电压进行了试验测量,结合相关文献的试验数据,验证了所提出的空气间隙起晕和击穿电压预测方法的有效性。最后,提出了一种提高长空气间隙绝缘强度的措施,并进行了试验验证。本文的研究成果对特高压输变电工程的电晕控制和空气间距选择具有一定的理论和工程应用价值。本文的主要内容及取得的成果如下： (1)提出了一种基于电场特征集和SVM的空气间隙起晕和击穿电压预测方法。提出采用电场特征集表征空气间隙结构,作为SVM的输入参量；以间隙是否起晕或击穿作为SVM的输出参量,即将起晕和击穿电压的预测由回归问题转换成二分类问题,建立了空气间隙起晕和击穿电压的预测模型。 (2)基于提出的方法,对棒-板间隙的正直流起晕电压进行了预测研究,通过对比预测值与试验值及已有方法的预测值,证明了所提出方法对起晕电压预测的有效性和优越性。提出了起晕电压测量的臭氧检测法,结合提出的预测方法,分析了绝缘罩对棒-板间隙负直流起晕电压的影响,结果表明：在棒-板间距大于一定值后,绝缘罩表面积累的空间电荷会提高棒-板间隙的负直流起晕电压值,起晕电压增益随间距的增大和绝缘罩内径的减小而增大。对+660kV直流换流站阀厅内均压球的起晕电压进行了试验和预测研究,得到了均压球表面的起晕场强控制值,通过对比电场有限元数值计算结果,证明了均压球在阀厅运行环境中不会起晕。 (3)将提出的方法应用于负直流下绞线的起晕电压预测中,将预测值与试验值及已有方法的预测值进行了对比。结合起晕电压的光电离预测模型,分析了绞线起晕电压与大气参数(气压、海拔、温度)和绞线结构参数(绞线半径、对地高度、分裂间距、分裂数)的关系,以及绞线表面粗糙系数与绞线半径和最外层细导线数的关系。 (4)采用提出的方法对稍不均匀电场短空气间隙的工频击穿电压进行了预测,同时开展了工频耐压试验,研究中考虑了球隙、棒-板和球-板等典型电极结构和球-板-球异形电极结构。通过对比预测值和试验值,验证了方法对稍不均匀电场短空气间隙击穿电压预测的有效性。对比分析了BP神经网络、RBF神经网络和SVM三种方法对考虑温湿度影响的球隙工频击穿电压的预测效果,证明了SVM方法在非线性逼近和泛化能力方面具有一定的优越性。 (5)针对极不均匀电场长空气间隙击穿特性明显受到电晕放电产生的空间电荷影响的问题,在稍不均匀电场短空气间隙击穿电压预测模型的基础上,提出了两种考虑电晕影响的方法,并进行了对比分析。将提出的方法成功应用于球-板长间隙的正极性操作冲击50%放电电压预测中,从而在一定精度要求范围内,可采用预测替代试验,达到减少试验次数和降低试验费用的效果。最后,提出了一种多间隙结构提高长空气间隙绝缘强度的措施,并进行了试验验证。
[Abstract]:UHV power grid can significantly improve the security, reliability, flexibility and economy of the power grid. The electromagnetic environment and external insulation characteristics are the key concerns of the design and operation of UHV transmission and transformation engineering. The core contents are corona control and air spacing selection, and the basic basis is the halo and breakdown characteristics of air gap, respectively. Nuclear test is the main basis for testing the requirements of UHV Engineering design, but the test has a long period and high cost. At the same time, it is difficult to exhaustion the actual transmission and Transformation Engineering gap. In addition, the research conclusions obtained by the experiment are mostly based on the gap scale, which can not be described by the geometric scale in the process. The gap structure has been greatly restricted. Therefore, it is necessary to carry out the research on the mechanism and model of air gap discharge, and predict the halo and breakdown voltage of various gap structures under different operating conditions by means of simulation, and use it to guide the design of actual transmission and transmission engineering.
At present, the prediction method of air clearance and breakdown voltage mainly includes experience, semi empirical formula and physical model. Experience, semi empirical formula, because some parameters are obtained under specific test conditions or premise assumptions, so the scope of application is limited. The physical model predicts corona or breakdown characteristics from the physical process of discharge development. It has made great progress in 20 years. However, because of the complexity of the physical model, there are still some unsolved problems, which lead to a large deviation between the predicted value and the test value, and it is difficult to effectively guide the actual engineering design. In this paper, an electric field feature set and a support vector machine (support vector machine) are proposed in this paper. SVM) the prediction method of air clearance and breakdown voltage. Based on the proposed method, the DC corona voltage and the breakdown voltage of different electrode structures in the transmission and transformation project are predicted, and the related factors are analyzed. At the same time, the halo and shock of some gap structures are also analyzed. The test measurements were carried out and the effectiveness of the proposed method for predicting the air clearance and breakdown voltage was verified by the experimental data of relevant literature. Finally, a measure to improve the insulation strength of the long air gap was proposed and tested. The results of the research on the corona control of the UHV transmission and transformation project. The selection of air distance has certain theoretical and engineering application value. The main contents and achievements of this paper are as follows:
(1) a prediction method of air clearance and breakdown voltage based on the characteristic set of electric field and SVM is proposed. It is proposed to use the characteristic set of the electric field to represent the air gap structure as the input parameter of the SVM. If the gap is halo or breakdown as the output parameter of the SVM, the prediction of the forthcoming halo and the breakdown voltage is converted from the regression to the two classification. A prediction model for the vignetting and breakdown voltage of the air gap is established.
(2) based on the proposed method, the positive DC corona voltage of the rod and plate gap is predicted. The effectiveness and superiority of the proposed method to the prediction of the halo voltage is proved by comparing the predicted values and the predicted values and the predicted values of the existing methods. The ozone detection method for the measurement of the halo voltage is proposed, and the analysis of the proposed method is analyzed. The effect of insulation cover on the negative DC corona voltage in rod and plate clearance is shown. The results show that after the rod and plate spacing is greater than a certain value, the space charge with the surface area of the insulating cover increases the negative DC corona voltage value of the rod and plate gap, and the gain of the halo voltage increases with the increase of the distance and the decrease of the inner diameter of the insulating cover. The valve hall of the +660kV DC converter station is increased. The corona voltage of the internal uniform pressure ball is tested and predicted, and the control value of the corona field strength on the surface of the pressure ball is obtained. By comparing the numerical results of the finite element analysis of the electric field, it is proved that the pressure sharing ball will not be halo in the operating environment of the valve hall.
(3) the proposed method is applied to the prediction of the halo voltage of the negative DC stranded wire. The predicted values are compared with the experimental values and the predicted values of the existing methods. Combined with the photoionization prediction model of the halo voltage, the stranded corona voltage and atmospheric parameters (air pressure, altitude, temperature) and the structure parameters of the strands (the radius of the twisted line, the ground height, and the height of the ground are analyzed. The relationship between the crack spacing, splitting number, and the surface roughness coefficient of the stranded wire is related to the radius of the stranded wire and the number of the outermost thin wires.
(4) the proposed method is used to predict the power frequency breakdown voltage of the short air gap in a slightly uneven electric field, and the power frequency pressure test is carried out at the same time. In the study, the typical electrode structure and the ball plate and ball special-shaped electrode structure are taken into consideration in the study. By comparing the predicted values and the experimental values, the method is proved to be slightly uneven electric field. The effectiveness of short air gap breakdown voltage prediction is compared and analyzed by comparing the prediction effect of three methods of BP neural network, RBF neural network and SVM on the frequency breakdown voltage of spherical gap considering the influence of temperature and humidity. It is proved that the SVM method has a certain advantage over the nonlinear approximation and generalization ability.
(5) on the basis of the prediction model of the short air gap breakdown voltage in a slightly uneven electric field, two methods to consider the effect of corona are put forward on the basis of the effect of the space charge of the corona discharge on the gap breakdown characteristic of the air gap in the extremely uneven electric field. The method is applied to the ball plate length successfully. The positive polar operation of the gap is used to predict the 50% discharge voltage. In the range of a certain precision, the prediction substitution test can be used to reduce the test times and reduce the cost of the test. Finally, a multi gap structure is proposed to improve the insulation strength of the long air gap, and the experimental verification is carried out.

【学位授予单位】：武汉大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TM83

【相似文献】