基于数据扩充和故障特征提取的ISAO-SVM断路器故障诊断研究

Research on fault diagnosis of ISAO-SVM circuit breaker based on data enrichment and fault feature extraction

摘要:

本研究提出一种断路器高效故障诊断方法。针对故障样本不均衡问题，采用合成少数类过采样技术(Synthetic Minority Oversampling Technique, SMOTE)进行样本扩充；采用蜣螂算法(Dung Beetle Optimizer, DBO)对变分模态分解(Variational Mode Decomposition, VMD)的惩罚因子、分解层数进行自适应选取；计算分解后各分量的熵值和时域指标，构建多维混合特征向量，并采用核主成分分析(Kernel Principal Component Analysis, KPCA)实现故障特征的提取；在支持向量机(Support Vector Machine, SVM)的超参数取值上，采用改进雪消融算法(Snow Ablation Optimizer, SAO)对SVM的参数进行寻优。将提取后的故障特征输入到SVM中进行故障诊断，实验结果表明，该模型能够较好的提取各样本的故障特征，具有良好的故障诊断效果。并将该模型与其他模型进行对比，该模型的诊断精准度均高于其他模型，具有较好的泛化能力。

Abstract:

The study proposes an efficient fault diagnosis method for circuit breakers. To address the issue of imbalanced fault samples, the Synthetic Minority Oversampling Technique (SMOTE) is used for sample expansion; Adopting the Dung Beetle Optimizer (DBO) algorithm to adaptively select the penalty factor and decomposition layers for Variational Mode Decomposition (VMD); Calculate the entropy values and time-domain indicators of each component after decomposition, and construct a multidimensional mixed feature vector, and use Kernel Principal Component Analysis (KPCA) to extract fault features; On the hyperparameter values of Support Vector Machine (SVM), an improved Snow Ablation Optimizer (SAO) algorithm is used to optimize the parameters of SVM. The extracted fault features are input into SVM for fault diagnosis. The experimental results show that the model can extract the fault features of each sample well and has a positive fault diagnosis effect. The model is compared with other models, and its diagnostic accuracy is higher than that of other models, which has better generalization ability.