摘要
为提高同步电机转子绕组匝间短路故障的诊断准确率,以一台型号为SDF-9的一对极同步发电机为研究对象,提出了一种基于二维卷积神经网络(Convolutional Neural Networks, CNN)与多源机电信息融合的匝间短路故障诊断方法。首先选取故障前后的定子环流、转子振动、定子振动信号为故障特征,采用信号-图像转换方法将一维时序信号转化为二维灰度图像。其次将处理后的图像分别作为二维CNN模型的前置输入进行训练。最后采用D-S证据理论将3种证据体的输出概率进行决策融合。结果表明:该方法消除了单一信号易受传感器故障及环境变化的影响,故障诊断准确率显著提高,并与其他传统故障诊断算法的诊断结果进行对比分析,验证了此方法的有效性。
In order to improve the diagnosis accuracy of interturn short circuit fault of synchronous motor rotor winding,we proposed an inter-turn short circuit fault diagnosis method based on two-dimensional Convolutional Neural Networks(CNN)and multi-source electromechanical information fusion with a one-pair synchronous generator of model SDF-9 as the research object.Firstly,we selected the stator circulation,rotor vibration and stator vibration signals before and after the fault as the fault features,and used the signal-to-image conversion method to convert the one-dimensional time-series signals into two-dimensional grayscale images.Secondly,we used the processed images as the pre-input of the two-dimensional CNN model for training.Finally,we used the D-S evidence theory to fuse the three evidence bodies for decision making.The results show that the method eliminates the single signal susceptible to sensor failure and environmental changes,and the fault diagnosis accuracy is significantly improved.In addition,the effectiveness of this method is verified by comparing and analyzing the diagnosis results with those of other traditional fault diagnosis algorithms.
作者
马明晗
侯岳佳
李永刚
贺鹏康
齐鹏
武玉才
MA Minghan;HOU Yuejia;LI Yonggang;HE Pengkang;QI Peng;WU Yucai(State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources(North China Electric Power University),Baoding 071003,China)
出处
《华北电力大学学报(自然科学版)》
CAS
北大核心
2024年第2期123-134,I0011,共13页
Journal of North China Electric Power University:Natural Science Edition
基金
国家自然科学基金资助项目(51777075,52177042)
河北省自然科学基金资助(E2020502064)
中央高校基本科研业务费专项资金资助项目(2021MS066)。
关键词
同步电机
转子绕组匝间短路故障
D-S证据组合
故障诊断
卷积神经网络
synchronous motors
rotor winding interturn short circuit fault
D-S evidence combination
fault diagnosis
convolutional neural network