LightGBM-ICOA-CNN Transformer Fault Diagnosis Method Based on DGA

To improve the accuracy of transformer fault diagnosis based on deep learning, a LightGBM-ICOA-CNN transformer fault diagnosis method based on dissolved gas analysis(DGA) is proposed. Firstly, the characteristic variables of transformer are enriched based on the content of dissolved gas in transformer oil, and then the light gradient boosting machine(LightGBM) is used to quantify their importance and achieve the feature variable preference. Secondly, the improved coati optimization algorithm(ICOA) is introduced to optimize the hyperparameters of convolutional neural network(CNN) such as learning rate, size and number of convolutional kernels, and number of fully connected layer neurons to improve the accuracy of the model diagnosis results. Finally, the performance of the established LightGBM-ICOA-CNN method is evaluated by an example analysis, which verifies the effectiveness of the proposed method for transformer fault diagnosis, and the convergence is good and the accuracy is high.