A physics-informed deep transfer reinforcement learning (PIDTRL) strategy is proposed for enabling advanced control and modulation of a dc solid state transformer (DCSST) in a dc microgrid. The strategy involves three stages: (i) Centralized training of deep reinforcement learning agents to balance power and reduce current stress in the DCSST; (ii) effective knowledge transfer from a Source-simulation system to a Target-experimental system using minimal experimental data; and (iii) deployment of multiple agents for online control in the DCSST. The proposed method adaptively determines optimal modulation variables (duty cycles and phase shifts) in stochastic and uncertain environments without requiring accurate model information. Experimental results validate the effectiveness of the proposed PIDTRL algorithm.

Full DC offshore wind power collection and transmission is an excellent scheme for long-distance offshore wind power development, in which wind turbine medium voltage DC/DC is one of the key technologies. A three-level series resonant switched-capacitor-active-fault interface DC/DC converter (TLSRSC-AFI) is proposed. Using switched-capacitor modules instead of full-bridge rectification to improve the boosting capability, which can reduce the number of input parallel output series (IPOS) power modules and reduce the size, weight, and cost. The active fault interface circuit operates normally when some sub-modules fail, thus improving the reliability of the system; at the same time, when the MVDC side fails, all AFI switches are turned off, so there is no DC path discharge current or no overcurrent, additionally, it has DC short-circuit fault blocking capability. Adopting a three-level structure, with automatic anti-bias capability, the switch tube voltage stress is reduced, and the power quality is high; adding LC series resonance to realize soft switching of switch tubes; input and output isolation, fault isolation. The steady-state operation, voltage relationship, current/voltage stress, and parameter design of the converter have been analyzed and derived in detail. The simulation was built in MATLAB and the experimental platform was designed to verify the validity of the theoretical analysis.