The invention discloses an intelligent identification method of a transformer station ground network for maximum step potential difference and maximum contact potential difference. The method is characterized by comprising steps that (1), ground potential rise and a surface potential function are calculated; (2), solution for the maximum step potential difference and the maximum contact potential difference is converted into a parameter optimization problem with constraints; (3), an artificial bee colony algorithm is employed to search optimal solutions of the maximum step potential difference and the maximum contact potential difference. The method is advantaged in that an identification problem of the maximum step potential difference and the maximum contact potential difference is converted into a derivative-less pure value optimization problem, the artificial bee colony algorithm is employed to search the optimal solutions, the employed method has properties of small computation complexity and fast computation speed, the maximum step potential difference and the maximum contact potential difference of a transformer station in the strict sense can be acquired, problems of relatively large computation complexity and non-high precision existing in a traditional observation surface equal interval sampling difference method are solved, and the method has great importance to safety evaluation on the transformer station ground network.