The invention discloses a big data statistical analysis based transformer substation factory boundary noise prediction and evaluation method, which is characterized in that information acquisition is performed on main equipment of a transformer substation, and core influencing factors influencing transformer substation factory boundary noise are acquired by using big data statistical analysis; data of a newly built/to-be-tested transformer substation is sent to an evaluation system to analyze through building a noise evaluation model in the evaluation system, and prediction and evaluation for the transformer substation factory boundary noise are realized by adopting a fuzzy evaluation method. The transformer substation factory boundary noise prediction and evaluation method provides reference proposals for transformer substation construction or renovation.