Uludağ Üniversitesi Mühendislik Fakültesi Dergisi

Öz

Flyback converter is one of the most preferred converter structures for low-power industrial applications. The input voltage and load, which are among the operating parameters of this converter, can vary, making the power control of the converter important. Various control techniques are used to control the flyback converter. The structures of these techniques can be simple or complex, while their implementations can be easy or difficult. Additionally, these techniques have advantages and disadvantages depending on whether they suit systems with linear, nonlinear, or changing operating parameters. In this study, a machine learning-based polynomial regression technique is proposed for the power control of the voltage-mode flyback converter, which has a nonlinear structure and dynamically changing operating parameters, such as input voltage and load. The control curve, which is required for the implementation of the proposed technique and shows the variation in switching frequency against error, is obtained for the reference value of 48V. The curve is modeled as a third-degree polynomial with the independent variable being the error and the dependent variable being the change in the switching frequency. The obtained polynomial is run within simple C blocks in the PSIM environment and used for the closed-loop power control of the converter with an output power of 154W. As a result of the simulation studies, it is observed that the output voltage of the polynomial regression-controlled flyback converter stably follows the reference voltage against the changing input voltage and load. Moreover, the power switch turns on under ZVS conditions in all operating states.