Abstract: Affected by the distribution of system impedance and the control method of the converter station,the bipolar DC grid generally does not operate near the optimal operating point during normal operation,which will cause the system to have large power loss and greatly reduce the ability to withstand external shocks. In response to the above problems,the idea of the simplex method and species "evolution mechanism" are introduced into the artificial fish school algorithm,and a new double fish swarm algorithm is proposed. The algorithm uses the simplex elite fishes to search in parallel with the ordinary fishes,and then the compound fish swarm is optimized according to a certain rule of "evolution". Finally,it is applied to the power flow optimization of the DC grid and the power loss and voltage imbalance are used as the optimization objectives. The results show that the algorithm can optimize the control parameters of droop control converter station in real time,so as to ensure that the whole system always runs near the best point,and ensure the security and economy of the DC power grid.