This paper proposes a 5G base station energy consumption optimization method based on federated learning. Through federated learning of base station data in different scenarios, accurate prediction of energy consumption in different scenarios is realized, so as to better optimize base station energy consumption. By collecting the base station service performance data, the cluster method is used to predict and judge the scene of the base station automatically. By applying federated learning to 5G base station energy consumption optimization, base station participants do not need to upload data, effectively solving data communication and data privacy issues.