The invention relates to a Dijlstra algorithm and genetic algorithm based power supply path optimization method, and aims at solving the problem that the efficiency of power generation equipment in astarting area is low due to the fact that a power supply path cannot be selected from the complex network reasonably or take practicality into consideration and a power plant cannot be connected to atransformer substation rapidly and effectively via the selected path. The method comprises the concrete steps that 1) a node and line model of a power system is established according to existing connection condition of a power system framework; 2) n nodes are set in the node and line model, a distance matrix of any two nodes is L, and a distance corresponding to a shortest path is the shortest distance LL between any two nodes in the node and line model; and 3) a node of a black start power supply is p, and a shortest path in which m of the n nodes are connected with the node p and a shortestdistance corresponding to the shortest path are obtained. The method is applied to the field of power system.