摘要
受到虚假数据干扰,导致高压开关柜负荷数据挖掘结果误差大,针对该问题,提出了基于深度学习的高压开关柜负荷数据智能挖掘系统设计。使用ARM+DSP双CPU结构,对适配器代理初始化,并采用双臂螺旋天线,设计负荷数据监测器,检测500~1500 MHz频带内的局放信号,抑制噪声干扰;通过CAN总线或485总线,将监控信息传送到智能交换机,实现远程监控;根据断路器接触点及电流特性,设计了电流互感器,使感应电压的变化范围变小;构建空间中连续一组函数MMD,调整原有网络结构,建立深度学习挖掘模型,初始网络参数,消除网络中虚假数据,利用目标域数据对网络优化,结合挖掘引擎实现数据智能挖掘。由实验结果可知,该系统挖掘误差为0,具有精准挖掘效果。
Due to the interference of false data,the error of high voltage switchgear load data mining results is large.Aiming at this problem,the design of high voltage switchgear load data intelligent mining system based on deep learning is proposed.ARM+DSP dual CPU structure is used to initialize the adapter agent,and the dual arm spiral antenna is used to design a load data monitor to detect the partial discharge signal in the 500~1500 MHz frequency band and suppress noise interference;The monitoring information is transmitted to intelligent switch through CAN bus or 485 bus to realize remote monitoring;current transformer is designed according to contact point of circuit breaker and current flow characteristics to reduce variation range of induced voltage;A group of continuous function MMD in space is constructed to adjust original network structure and establish deep learning mining model.It uses the target domain data to optimize the network,and combines with the mining engine to realize intelligent data mining.The experimental results show that the mining error of the system is 0,and it has accurate mining effect.
作者
万四维
廖肇毅
何俊达
WAN Siwei;LIAO Zhaoyi;HE Junda(Dongguan Power Supply Bureau of Guangdong Power Grid Co.,Ltd.,Dongguan 523000,China)
出处
《电子设计工程》
2022年第24期157-161,共5页
Electronic Design Engineering
关键词
深度学习
高压开关柜
负荷数据
智能挖掘
deep learning
high voltage switchgear
load data
intelligent mining