Research on Transformer Life Prediction Optimized by Artificial Intelligence Algorithm

To address the issues of low fitting accuracy and poor generalization capability in existing transformer life prediction models, this paper proposes a transformer life prediction method based on artificial intelligence algorithm optimization, aiming to improve prediction accuracy and stability. By constructing a transformer health index system, including the main health index, oil quality health index, and dissolved gas health index, and incorporating operational status and defect conditions as adjustment factors into the final health index. In this paper, a random forest lifetime prediction model based on the optimization of the Pelican algorithm (POA-RF) is developed. This approach resolves the parameter optimization problem in the Random Forest model. Through testing with actual operational data and comparison with other commonly used prediction algorithms, the effectiveness of the proposed method is validated. The training and testing results of transformer instances show that, compared to the unoptimized Random Forest model, the POA-RF model significantly reduces the mean absolute error (MAE), mean relative error (MAPE), mean square error (MSE), and root mean square error (RMSE), while improving the coefficient of determination (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>), demonstrating superior forecasting capability.