Data-driven PI control for DC–DC converters: A practical educational framework

This paper presents a practical data-driven adaptive PI control framework for DC–DC converters, integrating online Recursive Least Squares (RLS)-based system identification with automatic PI controller tuning to achieve robust voltage regulation under varying operating conditions and load types, including constant power loads (CPLs). The method bridges the gap between theoretical control design and practical implementation, making it suitable for both academic and industrial applications. One of the main challenges in controlling DC–DC converters is that conventional nonlinear techniques such as backstepping, sliding mode control, and high-order sliding mode control can handle uncertainties and disturbances but involve complex controller structures and parameter tuning procedures. These requirements often exceed the capabilities or interests of practical engineers and students, especially in educational or field-oriented environments. Moreover, in many power electronics applications, if the converter topology changes, the theoretical design must be reformulated from scratch, often requiring heavy mathematical derivations. In contrast, the proposed method requires only an identification of transfer function model and can be applied to various control objectives such as output voltage, input voltage, or output current regulation. Experimental tests on both buck and boost converters confirm that the proposed adaptive strategy can automatically retune the PI controller in real time, ensuring stable performance under varying load conditions. Compared to conventional fixed-gain PI tuning, the proposed method provides improved transient performance, with faster response and reduced settling time while maintaining satisfactory overshoot characteristics across the investigated buck and boost converter operating conditions. Owing to its simplicity, flexibility, and ease of implementation, the method is especially well-suited for instructional use in power electronics laboratories.