Proposed Reliability Prediction Model for Power Transformer Based on 2-FAL

Accurately estimating the remaining useful life (RUL) of power transformers is critical for maintaining the reliability and efficiency of electrical power systems. This study proposes a novel empirical model to estimate the degree of polymerization (DP) - a critical measure of insulation deterioration in transformers - by analyzing 2-furfuraldehyde (2-FAL) concentration. Developed using a dataset of 125 samples from published literature, the model employs logarithmic regression to capture the exponential relationship between 2-FAL accumulation and insulation deterioration. Validation through comparison with Chengdong's model and other existing approaches, with performance metrics including and R-square of 0.9917, mean absolute error (MAE) of 58.08, and root mean square error (RMSE) of 86.59. The model's predictions for 2-FAL concentration in ranges of 0–5 ppm and 0–40 ppm closely align with the measured DP values. More specifically, it reflects a rapid initial decline followed by gradual stabilization at higher concentrations. The high-precision DP estimation enhances RUL prediction accuracy, offering a robust foundation for data-driven maintenance strategies for power transformers.