基于LSTM神经网络的发电机绕组散热异常监测系统设计  

Design of Monitoring System for Abnormal Heat Dissipation of Generator Windings Based on LSTM Neural Network

作　　者：刘明哲 毛振攀 周桐 李锦云 乔加奇 LIU Ming-zhe;MAO Zhen-feng;ZHOU Tong;LI Jin-yun;QIAO Jia-qi(China Three GorgesRenewables(Group)Co.,Ltd.,Tongzhou District,Beijing 101199)

机构地区：[1]中国三峡新能源(集团)股份有限公司,北京101199

出　　处：《环境技术》2023年第7期69-74,共6页Environmental Technology

基　　金：中国三峡新能源(集团)股份有限公司科研项目资助(合同编号:三峡新能源合字[2021] 424号)。

摘　　要：随着电力能源需求的增加,发电机运转效率提高,增加了绕组散热量和热故障发生概率,影响发电机的正常运转。为此,设计了基于LSTM神经网络的发电机绕组散热异常监测系统。通过构造发电机绕组散热稳态平衡方程,分析发电机绕组散热特性,并结合绕组运行状态参数,构建发电机绕组温度模型。选取适当的绕组温度信号采集设备,实时采集发电机绕组温度信号,对其进行去噪处理,应用LSTM神经网络制定发电机绕组散热异常监测程序,实现了发电机绕组散热异常的检测与预警。实验结果表明,设计系统应用后,绕组温度信号信噪比最大值为97.41 dB,绕组散热异常监测结果与实际监测结果保持一致,证明了设计系统的应用性能良好。With the increasing demand for electric energy,the operating efficiency of generators increases,increasing the amount of heat dissipation from windings and the probability of thermal failure,affecting the normal operation of generators.Therefore,a monitoring system for abnormal heat dissipation of generator windings based on LSTM neural network is designed.By constructing a steadystate balance equation for heat dissipation of the generator winding,analyzing the heat dissipation characteristics of the generator winding,and combining the operating state parameters of the winding,a temperature model of the generator winding is constructed.Select an appropriate winding temperature signal acquisition device to collect the generator winding temperature signal in real time,perform noise removal processing on it,and apply LsTM neural network to develop a monitoring program for abnormal heat dissipation of the generator winding,achieving detection and early warning of abnormal heat dissipation of the generator winding.The experimental results show that after the application of the designed system,the maximum signal to noise ratio of the winding temperature signal is 97.4l dB,and the monitoring results of abnormal winding heat dissipation are consistent with the actual monitoring results,which proves that the application performance of the designed system is good.

关 键 词：发电机 异常监测系统 定子绕组热故障 运行状态 LSTM神经网络 绕组散热 

分 类 号：TM315[电气工程—电机]