Brain-Inspired Collaborative Power Dynamic Dispatch for the Digital Twin Smart Generation System

This paper proposes a brain-inspired collaborative power dynamic dispatch method for the digital twin smart generation system (DT-SGS) from the perspective of power dynamic dispatch. The physical power grid with large-scale renewable energy penetration can be divided into a ‘macro-micro’ double-layer architecture via the DT-SGS, for the purpose of low power generation cost and low carbon emission achievement. Due to the introduction of the digital twins, the operation mode of ‘offline training-online loading’ of the agent is determined to separate the online operation and agent training at the same time, thus avoiding the over-fitting phenomenon in the later stage of the agent learning. The brain-inspired collaborative power dynamic dispatch method is used to achieve the ‘centralized decision-decentralized control’ through the central agents so as to avoid being trapped into the local optima. Simulation experiments of the load frequency control system has been developed with the comparison of other control methods to prove the efficacy of the brain-inspired collaborative power dynamic dispatch method with lower power generation cost and carbon emission, better convergence and control accuracy.