Control and operation of DC microgrid

Abstract

As modern power systems incorporate increasing shares of renewable energy sources (RESs), the demand for flexible, efficient, and resilient power distribution architectures has grown accordingly. DC microgrids (DC MGs)—localized electrical grids that use direct current (DC) for power transmission and distribution—have attracted significant attention due to their reduced conversion losses, compatibility with DC-native sources and loads, and enhanced control performance. However, the intermittency of RESs, the destabilizing influence of constant power loads (CPLs), and harsh operational conditions, such as those encountered in marine environments, pose significant challenges to system stability. This thesis proposes robust nonlinear control architectures to address these challenges, aiming to maximize renewable energy harvesting while maintaining voltage stability under dynamic conditions. First, an in-depth review of DC MG architectures and practical applications, which lays the theoretical and contextual groundwork for the nonlinear control methodologies developed in the subsequent chapters. Second, a backstepping controller combined with a nonlinear disturbance observer (NDO) is developed for a novel PV-powered DC MG. This approach mitigates the negative impedance effects of CPLs and ensures maximum power point tracking (MPPT), even under variations in environmental irradiance and temperature. Based on maximum energy harvesting, the proposed controller performs a better dynamic response of bus voltage regulation compared to the traditional dual-PI controller. Finally, a compound large-signal stabilizer integrating model predictive control (MPC) with high-order NDOs is designed for interleaved multilevel boost DC-DC converters (IMBDCs) in shipboard microgrids (SMGs). The controller addresses parameter drift caused by environmental stresses, achieving rapid voltage regulation (settling time <10 ms) and inter-phase current balancing under large CPL variations, with superior performance over conventional dual-PI controllers. The proposed strategies are validated through simulations and real-time experiments, contributing to developing resilient, efficient DC MGs for critical applications such as marine electrification and sustainable data centers.