Electricity is essential in modern life, with consumption expected to rise by 80 % by 2024, making power transformers crucial. In developing countries, monitoring old-transformer power plants is often manually and infrequently, increasing damage and reducing transformer life. The lack of data limits the accuracy of machine learning, making traditional approaches less effective. This article introduces a new perspective through Cognitive Artificial Intelligence (CAI) with the Knowledge Growing System (KGS), which builds knowledge from scratch. KGS can detect and continuously learn about transformer degradation, improving predictive accuracy. This study demonstrates KGS’s ability to estimate transformer life while comparing its predictions with the Backpropagation Neural Network (BPNN) method. Enhancing decision-making in strategic planning ensures a reliable power supply and better transformer performance. It also supports the implementation of more intelligent and reliable preventive maintenance strategies. The method is as follows: • The KGS method demonstrates that the transformer is in satisfactory condition, with an estimated health level of 87.5 % in Semester 2 and 75 % in Semester 1. • The BPNN method estimates the transformer’s RUL at 23.42 years, achieving the RUL of 20.55 years or 7500 days with a normal loss of life of 0.0133 % per day