A Digital Voltage Trim Assisted Nonlinear Control Framework for Buck–Boost Converters Feeding Constant Power Loads

ABSTRACT

Conventional dc–dc converters employ pulse‐width modulation (PWM) for output‐voltage regulation, but finite PWM resolution at high switching frequencies and low output voltages introduces quantisation effects that degrade regulation accuracy and increase steady‐state ripple. This paper proposes a digital voltage trim control (DVTC)‐assisted nonlinear control framework for non‐inverting buck–boost converters supplying constant power loads (CPLs). A low‐bandwidth DAC‐based trim loop augments the primary PWM stage to improve voltage‐regulation resolution without modifying the internal switching architecture. A nonlinear integral backstepping controller (NI‐BSC) is developed to enhance transient regulation performance under the non‐minimum‐phase converter dynamics, supporting practically bounded closed‐loop behaviour under bounded disturbances and duty‐ratio saturation, while integral action reduces steady‐state regulation error. Controller gains are tuned offline via a reinforcement‐learning strategy based on prioritised experience replay (PER); only the trained actor policy is deployed at run time, preserving low online computational complexity. Experimental implementation on a CONTROLLINO MAXI platform with an MCP4725 DAC and ACS712 current sensor demonstrates reduced voltage ripple, improved transient recovery, and stable operation under reference variations, input‐voltage disturbances, and CPL changes. Simulation‐based benchmarking against RL‐assisted PI and predictive control approaches, together with experimental validation, confirms the effectiveness of the proposed framework.