ANALYSIS OF METHODS FOR MODELING POWER THREE WINDING TRANSFORMERS

This article presents a comparison of two approaches to modeling three-phase power transformers and the stages of further use of the model to create a digital twin of the product, including the application of neural network algorithms for classifying various modes of operation of an electrical apparatus. The first model was created using the standard Simulink block. The second model is developed on the basis of a T-shaped transformer replacement circuit, taking into account the active and inductive resistances of the windings and the magnetic circuit. The paper presents the corresponding limitations for the models being developed. The values of the magnetic circuit parameters are calculated based on the passport data of the physical object. To validate the model, acceptance tests of the transformer were carried out, namely, experiments on idling and short circuit. According to the data obtained, appropriate conclusions have been drawn about the correctness of using the presented computer model, which fully reflects all the physical phenomena of such an electrostatic device as a transformer and can be further used to train neural networks to solve a number of practical problems.