In DC microgrids, load power sharing and bus voltage regulation are two common control objectives. In this paper, a consensus-based algorithm is presented to achieve proportional power sharing and regulation of weighted geometric mean of bus voltages in DC microgrids with ZIP (constant impedance, constant current, and constant power) loads simultaneously. By using the virtue of the Laplacian matrices of undirected connected graphs, a lemma is derived to assist the stability analysis of the DC microgrids. Thus, a sufficient condition that stabilizes the system with ZIP loads is established. In addition, with the help of a distributed voltage regulation error estimator, the tuning of the bus voltages can be realized without the requirement on initial voltage conditions. Through Lyapunov synthesis, the large-signal stability of the closed-loop system is theoretically ensured for a wide range of load conditions. Finally, simulation studies are performed to validate the merits of the proposed consensus-based algorithm.

Direct current (dc) microgrids have been widely used in many critical applications. Such systems avoid unnecessary ac/dc conversions and can simplify control design. To achieve high-performance control of such system, advanced control algorithm needs to be designed. This paper presents a novel decentralized output constrained control algorithm for single-bus dc microgrids. The control objectives are to realize high-performance control of dc bus voltage, user-defined load sharing, and circulating current minimization. Unlike conventional control algorithms, the control algorithm can guarantee not only convergence but also bounded transient tracking error. By transforming the constrained system into an unconstrained system, the transient response of dc bus voltage can always stay within user-defined, time-varying bounds. Convergence of the transformed system can meet any transient performance requirement of the original system. Through proper control effort distribution, overall load demand can be shared among the distributed generators (DGs) according to predefined percentages. Accurate load sharing can indirectly minimize the harmful circulating currents among the DGs. Switch-level simulation and hardware experimentation with both single-bus and multiple-bus dc microgrids demonstrate the effectiveness of the proposed control design.

In modern control systems, most control algorithms, especially system-level ones, are implemented in discrete-time (DT) with digital controllers and digital communications. In this paper, a distributed DT algorithm is developed to achieve proportional load current sharing and average bus voltage regulation in DC microgrids. In order to reduce the communication requirement among the controllers, a periodic event-triggered (PET) DT algorithm is proposed by introducing a novel PET condition. Since the PET condition is detected periodically, the Zeno phenomenon existing in continuous-time (CT) event-triggered algorithms can be avoided. The communication burden for detecting the PET condition is also relieved since only the local and neighbors’ triggered information is required. Through the Lyapunov synthesis, the current sharing error is proved to converge to zero asymptotically. Finally, comparative results among different algorithms, based on a detailed switch-level microgrid model, are given to validate the effectiveness of the proposed PET control design.

In conventional DC microgrid control schemes, optimization and real-time control are usually performed at different time scales. It is hard for such control schemes to achieve real-time optimization. Even a small disturbance can result in deviations of bus voltages and output currents from their optimal operating points. In addition, most real-time controllers cannot guarantee the satisfaction of pre-defined constraints on individual bus voltages due to the disconnection between steady-state optimization and real-time control. In this paper, a distributed optimal control scheme is presented for DC microgrid. The objectives are to simultaneously realize generation cost minimization and individual bus voltage regulation. First, the optimal control problem is formulated, based on which a necessary and sufficient optimality condition is derived that characterizes the optimal operating points. The distributed optimal controller is then proposed to dynamically drive the system to operate at the optimality condition. Convergence to the optimal operating points that minimizes generation cost under constraints is guaranteed through rigorous Lyapunov synthesis. Each individual bus voltage can also be maintained within bounds in both transient-and steady-state. The performance of the proposed controller is validated through simulations based on a switch-level microgrid model.

Optimal generation allocation is customarily applied to the DC microgrids to enable optimal operation. Conventionally, the optimization is implemented periodically to obtain the optimal bus voltage or output power references for given operating conditions, which will unavoidably deviate from the actual ones. Since slight disturbances such as load changes will trigger real-time control adjustments, hence, the overall cost will increase due to the disconnect between optimization and real-time control. To overcome this issue, it is preferable to directly apply the optimal control method to render an optimal time path of control actions in real-time. This paper has studied the optimal generation control problem as a constrained non-convex problem with non-linearity. DRL has been successfully applied to solve such problems without mathematically modeling the actual system; the links between states and actions are discovered via ongoing environmental interactions, decreasing the reliance on system parameter information. This paper also showed that TD3-based optimal control could be applied to DC microgrids using a monotonical policy gradient search approach. Furthermore, DRL’s distributed training and execution framework is designed to realize real-time distributed control. The data sampling, storage, and experience buffer initialization strategy are customized to improve learning efficiency. The case study demonstrated its effectiveness.