DC/DC Boost Converter Control Using Levenberg-Marquardt (LM) Algorithm

In this paper, we discuss the application of a DC-DC boost converter control through Artificial Neural Networks (ANN). Traditionally, the small-signal linearized model near the operating point serves as the basis for the construction of the regular PI (proportional-integral) controller. As a result, the system's operating point model is altered, and the controller performance during starting, load changes, or reference changes is inadequate. By adhering to the standard model, the ANN is trained using Levenberg-Marquardt (LM) Algorithm to minimize the error function. The LM algorithm reduces the error of the system model with numerous parameters, which is a benefit of the ANN. The results of the simulation and experimental setup indicate that compared to conventional PI-based control the peak overshoot is reduced by 88.25% in ANN-based control of the boost converter.