基于深度学习的发电厂变频器过压故障检测  

Over⁃voltage fault detection of power plant inverter based on deep learning

作　　者：张宏伟 吕雪霞 ZHANG Hongwei;LV Xuexia(Tongliao Huolinhe Pithead Power Generration Co.,Ltd.,Tongliao 029200,China)

机构地区：[1]通辽霍林河坑口发电有限责任公司,内蒙古通辽029200

出　　处：《电子设计工程》2021年第5期71-74,79,共5页Electronic Design Engineering

基　　金：内蒙古自治区科学基金项目(2016MS0618)。

摘　　要：传统发电厂变频器过压故障检测方法信号提取能力差,当故障数据组过多时,检测能力有限,基于深度学习研究了一种新的发电厂变频器过压故障检测方法。设定初级深度学习结构,分别感知整流器感知器、逆变器感知器、电感与电容感知器是否存在故障因子,控制直流电源电路,提取故障信号。将故障信号转化成电信号,利用深度探究分析层寻找独特因子后分析逸散电子的稳定性,判定是否出现故障,并确定故障类型。实验结果表明,基于深度学习的发电厂变频器过压故障检测方法在解决多数据组故障问题时,解决能力优于传统的过压故障检测方法。The traditional detection method of inverter over⁃voltage fault in power plant has poor signal extraction ability.When there are too many fault data groups,the detection ability is limited.Based on deep learning,a new detection method of inverter over⁃voltage fault in power plant is studied.The primary deep learning structure is set to sense whether there is fault factor in rectifier sensor,inverter sensor,inductor and capacitor sensor respectively.The fault signal is transformed into electrical signal,and the stability of the escaping electron is analyzed by using the depth exploration analysis layer to find out the unique factors,to determine whether there is a fault,and to determine the fault type.The experimental results show that the method based on deep learning is better than the traditional method in solving the problem of multiple data sets.

关 键 词：深度学习 发电厂变频器 过压故障 故障因子 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]