Artificial Intelligence Based Switching Frequency Regulation with Fast Terminal Sliding Mode Control for DC-DC Step-Down Converters

Abstract

Making the best voltage/current regulation against disturbances and uncertainties and, most importantly, to minimize chattering effect in power electronics circuits, the hysteresis modulation based (HM) sliding mode control (SMC) is widely used. There are various research on this subject in the literature. However, the variable switching frequency that HM-SMC brought about has regrettably overlooked in the majority of these research. This problem is addressed in depth in this paper by demonstrating how the switching frequency is affected by various factors. Next, the fuzzy logic control (FLC) is proposed for the first time as a solution for switching frequency regulation in a power electronics circuit. The proposed FLC for switching frequency regulation fortifies the system described in this study and sets it apart from many other HM-SMC studies in the literature with its advantages against the difficulties caused by variable switching frequency such as excessive switching and driver power losses. On the other hand, the fast terminal SMC (FTSMC) which has significantly superior convergence performance than its counterparts, has been recommended for voltage regulation. The proposed FTSMC technique employs only one voltage sensor different from the past literature. This simplifies the proposed control mechanism in practical implementation and also makes the system economical. The two proposed controllers' compatibility with one another has been tested through experimental tests, including the steady-state condition, variations in switching frequency reference value, step change of load resistance, output voltage reference value variation, and finally step change of input voltage. The outcomes of the experiments demonstrate that the switching frequency and output voltage are successfully regulated with good dynamic and steady state performance. Finally, the proposed techniques have been compared with six recent approaches from the existing literature to highlight their performance.