Research Article
[Retracted] Research on Pattern Recognition Method of Transformer Partial Discharge Based on Artificial Neural Network
| Test items | Parameter setting |
| Convolution kernel size and step size | (5 ∗ 5, 1) | Pooling method | Max-pooling | Pooling window size and step size | (2 ∗ 2, 1) | Fill mode | Same mode | Activation function | ReLu | Weights | Random normal distribution | Number of LSTM nodes | 100 | Optimization function | SGD | Learning rate | 0.001 | Batch size | 64 | Epoch | 100 |
|
|