Identificação do conversor CC-CC boost usando modelos não-lineares de Hammerstein e Wiener

Abstract

Systems identification techniques act in modeling complex processes, which through representations and algorithms, estimate the parameters of an arrangement, which has or does not have a priori knowledge of its physical characteristics. The main contribution of this work is the parametric identification of a Boost converter by recursive estimation methods. For this, threeparametric representations were compared, one of which is the black box type described by the Autoregressive with Exogenous Input (ARX) linear model and two are the gray box type described by the Hammerstein and Wiener non-linear interconnected block models. Therefore, to acquire data from a real system to be identified, the structure and functioning of the step-up converter in continuous conduction mode are analyzed. The equations that describe the behavior of the Boost converter were developed to calculate the components that makeup such a circuit, which is based on the commercial characteristics of a photovoltaic module. To identify the parameters of the linear dynamic model, the duty cycle on a Pulse Width Modulation (PWM) applied to the transistor is excited by a Pseudo Random Binary Sequence (PRBS) to measure the output voltage, exciting the plant at different frequencies. The Recursive Least Squares(RLS) algorithm was used to estimate the ARX model, which also constitutes the structure of the block models, demonstrating that the number of iterations of this model results in parameters that satisfactorily define the dynamics of the system as their iterations increase. Simulations in MATLAB/SIMULINK are performed to determine the transfer function that relates the output voltage and the duty cycle of the converter. Finally, free simulation, step response, and closed-loop validation methods are implemented to validate the identified models.