Design and Development of a Transformer Condition Diagnosis System Application to Predict Purification Time and Life of a Transformer Based on Machine Learning

Transformers are equipment in high-voltage engineering that often experience abnormalities originating from internal or external sources. This abnormality causes disturbances, so it is necessary to diagnose faults in the transformer as early as possible so that fault predictions and maintenance actions can be carried out to maintain the quality of work on the transformer. An application was designed to diagnose the transformer's condition and predict purification time and transformer life to maintain the quality of the transformer's work. This application is designed to help make it easier to predict transformer conditions using dissolved gas analysis test results, oil scores, furan, and tap changer. The application design is done using the Python programming language and then converted into a mobile app using the React Native library framework with a simulator using Expo and connected to a database to store user data. The predicted data is for the GT 2.1 Excitation transformer, GT 2.1 Main Transformer, and GT 2.1 UAT transformer. Based on the application design, the test results were 100% successful; the system validated all the tests from the ten tests. The GT 2.1 Excitation transformer oil purification time is predicted to be 5697 days since the last test, and the transformer lifetime is predicted to be more than 15 years. The GT 2.1 Main Trafo transformer oil purification time is predicted to be 2697 days since the last test, and the transformer lifetime is predicted to be more than 15 years. The GT 2.1 Excitation transformer oil purification time is predicted to be 1866 days since the last test, and the transformer lifetime is predicted to be more than 15 years.