Active power dynamic distribution method for multi-energy power system

The invention discloses a multi-energy power system active power dynamic distribution method. The method belongs to the field of power dynamic distribution control among multiple power supplies in a power system. Aiming at the power dynamic distribution optimization problem of a wind, light and fire multi-energy power system, available power prediction of a wind turbine generator and a photovoltaic power station in the multi-energy power system is carried out by utilizing a GRU neural network, and renewable energy sources in the system can be reasonably utilized more effectively based on the change trend of meteorological conditions. In addition, by means of deep reinforcement learning, according to the control quality requirement of the multi-energy power system, scheduling strategies for units with different operation characteristics during dynamic power distribution are optimized accurately and rapidly at the same time. Meanwhile, the deep reinforcement learning performs optimization of a dynamic power dis