Nguyen-Widrow Neural Network for Distribution Transformer Lifetime Prediction

In the power system distribution, transformer has an important role to deliver electrical power to the consumer. The failure on the transformer could decrease the performance and remaining lifetime. Therefore, it is important to monitor the lifetime of the transformer to avoid the disturbance. This research proposes a simulation study to predict the lifetime of transformer using Nguyen-Widrow neural network. The parameters used for this research are current, temperature, and the lifetime of the transformer. Measurement is carried out on transformers with the rating of 20 kV/3S0-220 V and capacity of 100 kVA. The training and testing data of backpropagation neural network are PSD (power spectral density), and energy value resulted from the wavelet process. The expected output of the method is the prediction lifetime of the transformer. This study compares backpropagation methods. The results show that Nguyen-Widrow algorithm method can predict the lifetime of the transformer better than backpropagation neural network method.