A Fault Diagnosis Method of Power Transformer Based on Cost Sensitive One-Dimensional Convolution Neural Network

Machine learning based dissolved gas analysis (DGA) is a significant technique for the incipient fault diagnosis of power transformers. However, the diagnosis methods have limitations in learning from imbalanced fault dataset, especially in the case of a severe uneven distribution. The research interests, how to establish a suitable model for the imbalanced fault diagnosis, are drawing a great deal of attention. In this paper, a novel one-dimensional convolution neural network (1D CNN) model based on cost sensitive learning is proposed to be employed for transformer fault classification. Firstly, the structure of 1D CNN is designed with three hidden layers and two fully connected layers. Then, a class-dependent cost matrix is introduced into the soft-max function, which can modify the training process of the posed cost sensitive 1D CNN (CS-1D CNN), so as to pay more attention on the minority classes. Moreover, the PSO algorithm is adopted to optimize cost matrix for the CS-1D CNN. The performance of the proposed model is evaluated by case studies on a real-world fault dataset. The results reveal that the CS-1D CNN model has improved the accuracies of the minority fault classes and thus the overall accuracy of the whole dataset.