Parametric estimation in three-phase induction motors using torque data via the generalized normal distribution optimizer

Abstract

This research addresses the parametric estimation problem in three-phase induction motors by applying a recently developed metaheuristic method known as the generalized normal distribution optimizer (GNDO). A nonlinear programming model based on the steady-state circuit of the induction motor, which uses its Thevenin equivalent, is employed to model the estimation problem. Estimation is carried out by minimizing the mean square error between torque data (obtained from measurements or provided by the manufacturer) and the values calculated with the model. The main advantage of using the GNDO is its effective balance between the exploration and exploitation of the solution space via Gaussian distributions. Numerical tests in two three-phase induction machines confirm the effectiveness of this approach in comparison with the classical genetic algorithm, the particle swarm optimizer, and the sine cosine algorithm. The GNDO approach reports objective function values of about 9.7834×10−14 and 2.6500×10−14, while the sine cosine algorithm reaches solutions of about 4.6327×10−10 and 1.2400×10−6 in both tested motors. All numerical simulations were performed in the MATLAB software, version 2022b.