基于双向长短期记忆网络的气体绝缘全封闭组合电器设备的SF_(6)气体密度预测  

Study on SF_(6) Gas Density Prediction for GIS Equipment Based on Bi-LSTM

作　　者：戴丽莉[1] DAI Lili(Taiyuan Institute of Technology,Taiyuan,Shanxi 030001,China)

机构地区：[1]太原工业学院,山西太原030001

出　　处：《山西电力》2025年第1期15-19,共5页Shanxi Electric Power

摘　　要：气体绝缘全封闭组合电器设备中的SF_(6)密度是决定其绝缘和灭弧性能的关键因素,实际中采用带有温度补偿功能的气体压力表作为密度表监测气体绝缘全封闭组合电器设备的SF_(6)密度。由于正常运行时气室温度与环境温度并不平衡且压力表的温度补偿功能不可避免地存在一定误差,运维人员在巡视过程中很难凭借经验准确判断SF_(6)密度表读数的变化是否正常。基于此,结合历史运行数据,利用双向长短期记忆神经网络对气体绝缘全封闭组合电器设备SF_(6)密度进行了高精度预测,为现场工作人员准确判断气体绝缘全封闭组合电器设备运行状态提供了有力支撑。Since SF_(6) density of GIS equipment is a key determinant of its insulation and arc extinguishing performance,the gas pressure gauge with temperature compensation function is used as the density meter to monitor the SF_(6) density of GIS equipment in practice.Due to the imbalance between the air chamber temperature and the ambient temperature during normal operation,and the inevitable error in the temperature compensation function of the pressure gauge,it is difficult for operation and maintenance personnel to accurately determine whether the change of SF_(6) density meter reading is normal based on experience during inspection.In order to deal with this problem,Bi-LSTM neural network is used to predict SF_(6) density of GIS equipment with high precision based on historical operational data,which provides strong support for field staff to accurately determine the operating status of GIS equipment.

关 键 词：气体绝缘全封闭组合电器设备 SF_(6)密度 双向长短期记忆网络 时间序列预测 

分 类 号：TM595[电气工程—电器]