绝缘子沿面放电多光谱脉冲演化特性及诊断方法  

Multispectral Pulse Evolution Laws of Insulator Surface Discharges and Its Diagnosis Approach

作　　者：任明[1] 夏昌杰 余家赫 李信哲 董明[1] Ren Ming;Xia Changjie;Yu Jiahe;Li Xinzhe;Dong Ming(State Key Laboratory of Electrical Insulation and Power Equipment Xi’an Jiaotong University,Xi’an 710049 China)

机构地区：[1]电力设备电气绝缘国家重点实验室(西安交通大学),西安710049

出　　处：《电工技术学报》2023年第3期806-817,共12页Transactions of China Electrotechnical Society

基　　金：国家电网有限公司总部科技项目资助(5500-202218131A-1-1-ZN)。

摘　　要：绝缘子或套管处产生的沿面放电是导致开关柜、环网柜绝缘失效的常见诱因,而传统局部放电在线检测和诊断方法难以准确评估放电发展阶段和危害程度。该文采用单光子固态光电传感技术,对绝缘子沿面放电发展过程中的多光谱脉冲特性及其演化规律进行了试验研究,并提出一种基于多光谱脉冲演化特性的沿面放电严重程度诊断方法。首先,搭建多光谱光电同步检测平台,实现了沿面放电过程中多个光谱波段光信号和电流信号的同步采集;然后,根据全波段光脉冲幅值及其导数变化规律对沿面放电发展过程进行了三阶段划分,并分别对沿面放电三阶段多光谱脉冲的相位统计和非相位统计特征进行分析,明确了特征量随放电发展阶段的演化规律;最后,基于深度神经网络算法建立沿面放电多光谱严重程度评估模型,结果显示该模型诊断准确率达到96.75%,验证了多光谱诊断方法的可靠性。Insulator surface discharge is a common cause of insulation failure of switch cabinets.However,traditional surface discharge detection methods characterized by electromagnetic and acoustic signals can hardly meet the requirements in field application.Based on light radiation,another intrinsic nature of discharge,optical detection method has been applied for discharge diagnosis,performing strong anti-interference and high detection confidence.Further,spectral analysis provides a light wavelength information dimension for refined discharge diagnosis,but it is not practical because of the limitation of sensing equipment.In this paper,a practical discharge spectral detection sensor named SiPM-based multispectral discharge sensor(SMDS)is introduced to study the optical radiation characteristics and the evolution laws of multispectral characteristics during the surface discharge development process of insulators,moreover,a strategy for severity diagnosis of surface discharge based on the characteristics of multispectral pulse evolution are proposed.First,based on the new generation solid state photoelectric sensors,SMDS with seven specific band channels and one full band channel are adopted to acquire multispectral pulses of surface discharge.Second,the evolution stages of surface discharge are distinguished into three stages,named slight,moderate and severe stage,respectively,according to the derivative of full-band light pulse intensity at different voltages.Third,multispectral characteristics of surface discharge at different stages are analyzed.For the phase-based multispectral characteristics,the light intensity of each band increases significantly with the applied voltage independent of the phase,while the light intensity proportion of each band shows different evolution trends.In the slight stage,the light intensity at 450nm is the largest,accounting for about half of the total light intensity.While the light intensity of 400nm band has the smallest proportion.In the moderate stage,the proportion of light intens

关 键 词：沿面放电 光测法 多光谱诊断 深度神经网络 严重程度评估模型 

分 类 号：TM85[电气工程—高电压与绝缘技术]