Facing mountains of data in modern energy management system, Related operators need to use machine learning to derive corresponding knowledge to support its decision-making. In view of the above question, Based on Prism, FURIA and the J48 classifier, this paper used 10 fold cross validation on the energy management system for training a data table TP rate respectively were: 92%, 88% and 84%, Prism classifier produced 5 rules, FURIA classifier produced 4 rules, decision tree generated by J48 had 5 valid braches ∘ Rules generated by classifiers can provide decision-making guidance for energy management system, and accelerate decision-making response performance.