基于SMOTE-SSA-CNN的开关柜故障诊断方法  

Fault Diagnosis Method for Switchgear Based on SMOTE-SSA-CNN

作　　者：张玮 ZHANG Wei(State Grid Gannan Power Supply Company,Gannan Tibetan Autonomous Prefecture 747000,Gansu,China)

机构地区：[1]国网甘南供电公司,甘肃甘南藏族自治州747000

出　　处：《电气传动》2024年第10期83-89,共7页Electric Drive

基　　金：甘肃省重点研发计划(21YF5GA159)。

摘　　要：开关柜多源监测数据包含丰富的设备运行状态信息,对其进行分析可实现开关柜故障诊断。提出一种基于SMOTE-SSA-CNN的开关柜故障诊断方法。首先,以开关柜电压、电流和温湿度等监测数据为基础,采用合成少数类样本过采样技术(SMOTE)算法对原始数据集进行样本扩充,解决原始数据集中正负样本严重失衡的问题;然后引入麻雀搜索算法(SSA)对卷积神经网络(CNN)的卷积核大小与数量、全连接层神经元数量、学习率等超参数进行优化,提高模型故障诊断结果的准确率;最后,通过算例分析对建立的SMOTE-SSA-CNN模型性能进行评估,验证了所提方法对开关柜故障诊断的有效性,且与传统故障诊断方法相比,所提方法的收敛性较好,精度较高。The multi-source monitoring data of switchgear contains rich equipment operating status information,and analyzing it can achieve switchgear fault diagnosis.A fault diagnosis method for switchgear based on SMOTE-SSA-CNN was proposed.Firstly,based on monitoring data such as switchgear voltage,current,and temperature and humidity,the synthetic minority over-sampling technique(SMOTE)algorithm was used to expand the original dataset,solving the problem of severe imbalance between positive and negative samples in the original dataset.Then,the sparrow search algorithm(SSA)was introduced to optimize the hyperparameters of convolutional neural networks(CNN),such as the size and number of convolutional kernels,the number of fully connected layer neurons,and the learning rate,in order to improve the accuracy of the model's fault diagnosis results.Finally,the performance of the established SMOTE-SSA-CNN model was evaluated through example analysis,verifying the effectiveness of the proposed method for switchgear fault diagnosis.Compared with traditional fault diagnosis methods,the proposed method has better convergence and higher accuracy.

关 键 词：开关柜 多源监测数据 合成少数类样本过采样技术算法 麻雀搜索算法 卷积神经网络 

分 类 号：TM591[电气工程—电器]