Power Transformer Health Index and Life Span Prediction Using Supervised Machine Learning: A Hybrid Autoencoder Enhanced Gradient Boosting Approach

Health Index (HI) assesment serves as a vital indicator for ensuring reliability and evaluating operational state of power transformers. However, condition monitoring systems often operate under constraints of computational efficiency and latency, necessitating compact yet informative feature representations. To solve this problem, this paper devised a hybrid model that jointly optimizes the extraction of deep features through autoencoders via Gradient Boosting Regressor (GBR) to estimate transformer HI and lifespan resource efficiently without necessarily suffering accuracy loss. The proposed Autoencoder-GBR model is subsequently bench marked with the traditional feature reduction algorithms and the full uncompressed feature set together with the traditional as well as the deep learning based regression algorithms. The comparative analyses demonstrates that the recommended model has been able to perform better with an averageR2performance of 0.993 and an RMSE of 1.466 with dimensionality of features cut down by 35.7% that calls on both predictive ability and computational efficiency. The proposed model has less inference latency and has an average model size with insignificant peak memory. It also shows strong robustness with acceptable predictive performance under synthetic noise and missing feature scenarios hence ensures runtime feasibility and reliability for real world monitoring environment.Then the same Autoencoder-GBR model is applied for lifespan prediction which achieves anR2of 0.9946 and RMSE of 1.3022, further confirming its robustness. Overall, these findings demonstrates that the nonlinear feature compression not only boosts prediction accuracy but also ensures scalability, making proposed framework highly suitable for real time and field deployable transformer health monitoring systems.