Estimação de parâmetros de transformadores de potência em regime permanente utilizando o algoritmo Mean-Variance Mapping Optimization (MVMO)

Abstract

This undergraduate thesis presents the application of the Mean- Variance Mapping Optimization (MVMO) algorithm for estimating electrical parameters of power transformers under steady-state operation. Accurate identification of these parameters is essential for adequate representation of electrical behavior in performance studies, loss analysis, protection schemes, and power system reliability assessments. The proposed methodology is based on the modeling of the single-phase equivalent circuit referred to the primary side, incorporating resistances, series reactances, and magnetizing branch parameters. Parameter estimation is formulated as an optimization problem in which MVMO minimizes the error between measured and calculated electrical quantities from operational data. The method is evaluated on three transformers with different power ratings (2 kVA, 15 kVA, and 33 MVA), comparing its performance with five well- established algorithms: Genetic Algorithm, Particle Swarm Optimization, Gravitational Search Algorithm, Bacterial Foraging Algorithm, and Caothic Optimization Algorithm. Results demonstrate good performance of MVMO, with parameter errors lower than 0.2% for the 15 kVA and 33 MVA cases, establishing it as a viable alternative for parametric identification of transformers across different power scales.