Novel power system multi-target robust optimization method based on element multi-agent deep reinforcement learning

The invention discloses a novel power system multi-target robust optimization method based on element multi-agent deep reinforcement learning, and relates to the field of optimization of new energy power generation access power grids such as wind power, photovoltaic power and the like. The method specifically comprises the following steps: firstly, modeling a multi-target two-stage robust optimization mathematical model of a wind power/photovoltaic/thermal power hybrid energy system, and then solving a two-stage robust optimization problem through multi-agent reinforcement learning and solving a multi-target optimization problem through element reinforcement learning. And finally, combining the two methods to form a multi-target two-stage robust optimization end-to-end solution for the wind power/photovoltaic/thermal power hybrid energy system. The method is used for solving the multi-target two-stage robust optimization problem of the new energy power generation access power grid, and more efficient, more fl