In response to the insufficient correlation between traditional network captured acoustic features of On-load tap chang-er(OLTC),leading to low accuracy in fault diagnosis,this study proposes a fault diagnosis method for OLTC based on Transformer neural network.Firstly,Mel-frequency cepstral coefficients(MFCC)are utilized to extract acoustic features from OLTC sound sam-ples,reducing the data dimensionality of the samples.Subsequently,the Transformer is employed to comprehensively capture the re-lationships among acoustic features and achieve fault diagnosis for OLTC.Experimental results demonstrate that the Transformer-based approach achieves a high accuracy rate of 97.5％in diagnosing faults such as transmission shaft looseness,contact wear,stick-ing,and gear engagement issues in OLTC.Moreover,it contributes to a certain extent in shortening the diagnostic time.
关键词
有载分接开关/声纹特征/故障诊断/梅尔频率倒谱系数/Transformer
Key words
on-load tap changer/acoustic features/fault diagnosis/mel-frequency cepstral coefficients/transformer
维普
万方数据