With the global climate change increasingly urgent

the innovative energy dispatch is crucial for energy saving and carbon reduction. Mobile energy storage

with its temporal-spatial flexibility

can effectively promote the low-carbon energy use and enhance the activity level of carbon trading markets. In order to optimize the temporal-spatial dispatch of mobile energy storage in the electricity and carbon markets

a method based on two-layer multi-agent deep reinforcement learning framework is proposed in this paper. Firstly

an optimized dispatch model for mobile energy storage is established

which considers ladder-type carbon trading costs

spatial transfer costs

capacity decay costs

and charge-discharge arbitrage benefits. Secondly

the dispatch problem is formulated as a Markov game model

and a two-layer multi-agent deep reinforcement learning framework is constructed to solve the proposed model. Finally

the locational marginal price data and the location information of 30 charging stations in San Diego

California

from 2020 to 2022 is trained and simulated in the model. The results demonstrate that the proposed method can facilitate applicability

stability

and scalability and realize energy-saving and carbon reduction in the temporal-spatial dispatch process of mobile energy storage.