Modified Convolutional Neural Network-based Adaptive Fault Identification for Transformer On-load Tap Changer
The conventional adaptive identification of on-load tap changer faults in transformers often uses improved semi supervised ladder network algorithms,but due to the inability to solve the problem of network gradient explosion,the fi-nal fault identification accuracy is low.To this end,this work studied anomaly detection schemes for switchgear.Wavelet packet decomposition algorithm and frequency domain recognition vector of signal are used to mine the fault characteristic parameters.Utilizing conventional networks,features were integrated and network structures were updated.Then a fault adaptive recognition model was constructed,and the gradient explosion problem of the network structure was solved through residual structure.The comprehensive fault score of the input sample was obtained to determine the type of fault the sample,thereby achieving adaptive fault recognition.The example application results showed that the proposed method can effectively identify the fault of the on-load tap-changer,and the identification results are consistent with the actual results,and the highest F1_score value is 0.97,which has a high identification accuracy.
modified convolutional neural networkstransformer on load tap changerfault adaptive identificationidentification accuracy
万方数据
维普
