【作者】 周游；

【导师】 刘凯； 吴科；

【作者基本信息】 四川大学 ， 电气工程（专业学位）， 2021， 硕士

【摘要】 电力系统配电网接地运行方式有两种,一种是中性点经消弧线圈接地方式(Neutral Point Grounded System Through Arc Suppression Coil,简称NES),另一种是中性点不接地方式(Neutral Point Ungrounded System,简称NUS)。由于国民经济发展不断提速,电网运行需要具备更高的安全性能。在配电网的实际运行中,单相接地故障的发生概率是最高的,而发生故障时,目前多数供电企业依然采用拉路查找的方法来判断故障线路,该方法不但效率低下,而且会使非故障线路短时间断电,所以研究单相接地故障的选线、故障测距、故障定位等成为了配电网安全可靠性提升中非常重要的研究方向,具有较高的理论意义与广阔的工程应用价值。本文选择研究对象为配电网单相接地故障的选线,通过对配电网单相接地故障选线的原理进行深入研究与分析后,对其暂态特性及稳态特性进行了详细阐述。以单相接地故障暂态特性为基础,结合小波分析来实现配电网单相接地故障选线。仿真结果表明小波分析对非有效中性点接地系统具有的一定的选线性能。以此为基础,进行了算法理论升级,使用小波神经网络、粒子群优化小波神经网络以及经验小波分析依次对单相接地故障进行了选线分析,仿真结果表明算法融合度樾高,识别正确率也会越高。本文开展了以下的主要研究内容:(1)对配电网基本的单相接地故障选线原理进行了阐述,并分析了其暂态特性与稳态特性,着重说明了单相接地故障发生后,配电网零序参数变化情况;(2)结合小波分析实现配电网单相接地故障选线工作,并说明了其选线基本原理与流程,即通过模极大值以及极性变化来判断故障线路,从仿真结果看,小波分析对单相故障有一定的选线效率;(3)采用了基于小波神经网络与故障测度的融合算法,将其应用于单相接地故障选线中。相较于小波分析,小波神经网络的故障选线正确率有了较大提高;(4)采用粒子群优化算法为基础的小波神经网络进行单相接地故障选线,对其基本原理和实现思路进行了阐述。相较于小波神经网络,粒子群算法优化后的故障选线正确率有了小幅提高。(5)简明阐述了EWT算法的基本原理,形成了基于低频分量相关系数与高频分量能量权重系数的逻辑“或”判断依据用于选线。更多还原

【Abstract】 There are two grounding operation modes of power System.One is Neutral Point Grounded System Through Arc Suppression Coil(NES).The other is the Neutral Point Ungrounded System(NUS).As the national economy develops faster and faster,the power grid operation needs to have higher safety performance.In the actual operation of the distribution network,the occurrence probability of single-phase ground fault is the highest.When the fault occurs,most power supply enterprises still use the pull circuit search method to judge the fault line.This method is not only inefficient,but also makes the non-fault line temporarily disconnected.Therefore,the research on line selection,fault location and fault location of single-phase grounding fault has become a very important research direction in the safety and reliability improvement of distribution network,which has high theoretical significance and broad engineering application value.In this thesis,the research object is single-phase grounding fault line selection of distribution network.Through in-depth study and analysis of the principle of single-phase grounding fault line selection of distribution network,the transient characteristics and steady-state characteristics are elaborated in detail.Based on the transient characteristics of single-phase grounding fault,combined with wavelet analysis,line selection of single-phase grounding fault in distribution network is realized.The simulation results show that the wavelet analysis has a certain performance of line selection for the non-effective neutral grounding system.Based on this,the theory of the algorithm is upgraded.The wavelet neural network,particle swarm optimization wavelet neural network and empirical wavelet analysis are used to conduct line selection analysis for single-phase grounding fault.The simulation results show that the higher the fusion degree of the algorithm is,the higher the recognition accuracy will be.This thesis has carried out the following main research contents:(1)The basic principle of single-phase grounding fault line selection of distribution network is expounded,and its transient and steady characteristics are analyzed.The change of zero-sequence parameters of distribution network after single-phase grounding fault occurs is emphatically explained.(2)The line selection of single-phase grounding fault in distribution network is realized by combining wavelet analysis,and the basic principle and process of line selection are illustrated,that is,the fault lines are judged by the mode maximum value and polarity change.From the simulation results,the wavelet analysis has certain line selection efficiency for single-phase fault.(3)A fusion algorithm based on wavelet neural network and fault measure is adopted and applied to single-phase grounding fault line selection.Compared with the wavelet analysis,the fault line selection accuracy of the wavelet neural network is greatly improved.(4)The wavelet neural network based on particle swarm optimization algorithm is used for single-phase grounding fault line selection,and its basic principle and implementation idea are described.Compared with the wavelet neural network,the accuracy of fault line selection improved slightly after Pso optimization.(5)The basic principle of EWT algorithm is briefly explained,and a logical "OR" judgment basis based on the correlation coefficient of low frequency component and the energy weight coefficient of high frequency component is formed for line selection.更多还原