【作者】 刘丹；

【导师】 谢庆；

【作者基本信息】 华北电力大学 ， 电工理论与新技术， 2017， 硕士

【摘要】 电力变压器内部发生局部放电时,其发生位置对电力变压器绝缘状况的评估影响很大。现有的基于MUSIC的局放超声检测方法,其测向精度在一定程度上,满足工程实际的要求。但是,由于MUSIC算法,采用的是二维空间谱峰搜索,其等高线图以及谱峰图不能直观的从图中看出局放源的波达方向(DOA)估计量,且计算复杂性较高,测向精度较差,还有待进一步提高。据此,论文重点开展了降维MUSIC算法在局放超声阵列检测中的应用研究。主要内容如下:(1)提出了一种降维MUSIC的局放超声阵列的测向方法。该方法将局放超声阵列信号的波达方向(方位角和俯仰角)的谱峰空间二维并行搜索方式,降维分解为方位角和俯仰角的一维顺序搜索方式。建立了局放信号和超声阵列传感器的仿真模型,研究了降维MUSIC算法的性能以及复杂度,开展了基于降维MUSIC的局放超声阵列对单局放源及双局放源的仿真研究,并与基于Fast DOA算法和Root-MUSIC算法的局放超声阵列对单局放源仿真结果进行了对比分析。结果表明:该方法计算复杂性较低,计算速度快,测向精度高,分辨性能好,并且其测向星谱图可直观估计出信号源的方位信息,验证了该测向方法对局放源DOA估计的可行性。(2)开展了基于降维MUSIC的局放超声稀疏阵列传感器对局放源进行定位的仿真研究。采用降维MUSIC算法,利用4种稀疏阵列传感器对单局放源及双局放源进行定位仿真研究。仿真结果表明:测向性能较好的降维MUSIC算法,可以弥补阵列结构本身的测向性能的不足,降低了阵列结构对测向误差的影响,进而提高定位准确度。(3)完善了实验平台,开展了将降维MUSIC算法应用于处理满阵及稀疏阵列分别采集的模拟实验中实测的局放超声阵列信号的实验研究。实验结果验证了基于降维MUSIC测向方法的满阵及稀疏阵列对单局放源及双局放源均可实现DOA的估计,并且其测向精度较高,进而提高了下一步定位的准确度,为实际应用提供重要的指导信息。更多还原

【Abstract】 When the partial discharge(PD)occurs in the power transformer,the location of the PD has a great influence on the evaluation of the insulation condition of the power transformer.The direction finding accuracy of the PD detection based on the MUSIC algorithm meets the actual requirements of the project.However,due to the MUSIC algorithm,using a two-dimensional spatial spectrum peak search,the contour map and the peak figure cann’t be visually seen the PD source direction of arrival(DOA)estimation from the figures,and it has high computational complexity,poor accuracy,remains to be further improved.This paper is emphatically to carry out the research on the application of reduced dimension MUSIC(RD-MUSIC)algorithm in the PD ultrasonic array detection.The main contents are as follows:(1)Direction finding algorithm of the PD ultrasonic array based on the RD-MUSIC is proposed.The algorithm that made two dimensional parallel search method to be decomposed into the one-dimensional sequence search of azimuth and elevation angle.A simulation model of the PD signal and the ultrasonic array sensor was established.And the performance and complexity of the RD-MUSIC algorithm were studied.Moreover,the simulations of the single source and double sources were carried out based on the RD-MUSIC,and compared with the FastDOA algorithm and the Root-MUSIC algorithm.The results show that the RD-MUSIC algorithm has low computational complexity,fast calculation speed,high direction accuracy,and the resolution are good,besides,the star spectrum can estimate the azimuth information about the signal source directely.Finally,the simulations verify the feasibility of the algorithm.(2)The simulation of the sparse array sensor based on the RD-MUSIC algorithm was carried out.Using the RD-MUSIC algorithm,the simulations of single source and double sources with 4 sparse array sensors were studied.The simulation results show that the RD-MUSIC algorithm with better direction finding performance can make up the deficiency of the direction finding performance of the array structure itself,and reduce the influence of the array structure on the direction finding error,and thus improve the positioning accuracy.(3)The experimental platform was improved,the RD-MUSIC algorithm was used to deal with the measured PD ultrasonic array signal,which was collected with the full array and sparse array respectively.Moreover,the experimental results verify that the full array and sparse array based on RD-MUSIC method can be used to realize the estimation DOA for single source and double PD sources.It can realize the accurate direction finding of the sources,improve the positioning accuracy,and provide the important guidance information in the practical application.更多还原