A modeling framework for dc microgrids and distribution systems based on the dual active bridge (DAB) topology is presented. The purpose of this framework is to accurately characterize dynamic behavior of multi-converter systems as a function of exogenous load and source inputs. The base model is derived for deterministic inputs and then extended for the case of stochastic load behavior. At the core of the modeling framework is a large-signal DAB model that accurately describes the dynamics of both ac and dc state variables. This model addresses limitations of existing DAB converter models, which are not suitable for system-level analysis due to inaccuracy and poor upward scalability. The converter model acts as a fundamental building block in a general procedure for constructing models of multi-converter systems. System-level model construction is only possible due to structural properties of the converter model that mitigate prohibitive increases in size and complexity.

To characterize the impact of randomness in practical loads, stochastic load descriptions are included in the deterministic dynamic model. The combined behavior of distributed loads is represented by a continuous-time stochastic process. Models that govern this load process are generated using a new modeling procedure, which builds incrementally from individual device-level representations. To merge the stochastic load process and deterministic dynamic models, the microgrid is modeled as a stochastic hybrid system. The stochastic hybrid model predicts the evolution of moments of dynamic state variables as a function of load model parameters. Moments of dynamic states provide useful approximations of typical system operating conditions over time. Applications of the deterministic models include system stability analysis and computationally efficient time-domain simulation. The stochastic hybrid models provide a framework for performance assessment and optimization.