Method for Diagnosing Line Fault Arcs in Three-Phase AC Motors Under Vibration Conditions
Arc fault is an air breakdown phenomenon caused by loose electrical connections or aging insulation in wiring or equipment,which can lead to equipment damage,circuit faults,and even electrical fires.To accurately identify three-phase AC motor vibration fault arcs,a fault diagnosis model based on one-dimensional convolutional neural networks(1D CNN)combined with the pelican optimization algorithm(POA)is proposed.The proposed model can be trained directly using one-dimensional temporal current data output from the host computer.The POA is utilized to find the optimal hyperparameters of the 1D CNN,enhancing the model's recognition capability.After analyzing the feature extraction effectiveness of the model using the t-SNE algorithm,its validity is confirmed.Test results show that the fault arc identification accuracy reaches 99.72％.The proposed method is more convenient and superior in performance compared to existing technologies.
convolutional neural network(CNN)three-phase AC arc faultpelican optimization algorithm(POA)fault diagnosis
万方数据
维普
