A Comprehensive Design With Comparative Analysis of ANN-Based STSMC and Optimized PI Compensator for DC-DC Converter Feeding CPLs

In this article, a neural network based on the Super-Twisting sliding mode control (STSMC) is proposed for controlling the voltage of a DC-DC converter in an embedded propulsion system under constant power load (CPL). The converter model is then linearized in the neighborhood of the equilibrium point method, and optimal parameters for the PI compensator in various local operating regions are determined. Data collected through the STSMC approach are used as features and targets for training an Artificial Neural Network (ANN). An adjustable STSMC with tuned parameters supplied by ANN is tested by numerical simulations on a DC-DC converter, specifically a boost converter. The results demonstrate that the ANN-based approach offers superior dynamic voltage response compared to both the STSMC controller and conventional PI controller, with shorter stabilization times and minimal overshoot.