Muti-physics cosonstrained PEV charging management for solar PV penetrated energy-transportation Nexus: A novel DRL-based approach combining a cyber-physical system

Utilizing energy storage capability of plug-in electric vehicles (PEVs) to integrate the fluctuating surplus energy from increasingly penetrated solar photovoltaics (PVs) is an ideal solution to the instability in distribution systems. However, high rate charging regardless of the physical limits may induce irreversible degradation and even catastrophic safety issues. Regarding this, this paper proposes a data-driven charging management strategy with multi-physical awareness within a cyber-physical fusion framework. In particular, an electro-thermal model is established in the cyber layer, leveraging which a soft actor-critic strategy is innovatively exploited to make an intelligent balance over conflicting objectives and virtually optimize the power allocation with accelerated iterative convergence. The trained strategy is further downloaded for real-time PEV charging control in the physical layer. Hardware-in-Loop tests and experiments with actual LIBs are carried out for verification, suggesting that the proposed strategy can effectively mitigate the surplus power from solar PVs, with avoiding violation of the electrical and thermal safety limits of LIBs. Meanwhile, the proposed strategy is proven superior to the rule-based charging modes in terms of efficiency, charging safety, energy usage cost and computational complexity, with the trained low-complexity strategy highly adaptive to the ambient temperature, initial charging state and the states of distribution system as well, promising robust performance in practical applications.