摘要
为了识别开关柜内的绝缘缺陷,设计了绝缘缺陷实验模型以获取训练样本,建立了具有自学习能力的BP(back propagation)神经网络并引入遗传算法进行网络优化,神经网络输入层为开关柜内气体特征,输出层为开关柜内绝缘缺陷,建立了开关柜内气体特征与绝缘缺陷的映射关系。经过样本训练,优化后的神经网络能够准确地识别绝缘缺陷。
In order to distinguish the insulation defect in switchgear, models of insulation defects are designed to obtain training samples, BP(back propagation) neural network with self-learning ability is established and genetic algorithm is used to optimize network. The input layer of neural network is the gas characteristic in switchgear,and the outer layer is insulation defects in switchgear. The mapping relationship between gas characteristics and insulation defects in switchgear is established. After samples training, the neural network can reliably distinguish the types of insulation defects and realize the insulation defects monitoring of air switchgear.
作者
张振宇
晋涛
王天正
邵云峰
刘永强
王鹏
ZHANG Zhenyu;JIN Tao;WANG Tianzheng;SHAO Yunfeng;LIU Yongqiang;WANG Peng(Electric Power Research Institute,State Grid Shanxi Electric Power Company,Taiyuan 030001,China;State Grid Lüliang Power Supply Company,Lüliang 030000,China;School of Electrical Engineering and Automation,Wuhan University,Wuhan 430072,China)
出处
《武汉大学学报(工学版)》
CAS
CSCD
北大核心
2021年第8期761-768,共8页
Engineering Journal of Wuhan University
关键词
空气开关柜
局部放电
气体检测
BP神经网络
绝缘缺陷识别
air switchgear
partial discharge
gas detection
BP neural network
insulation defects recognition