There is an increasing need for reducing greenhouse gas emissions and improving the efficiency of energy consumption. In particular, energy consumption in the domestic industrial sector accounts for a large portion of the total energy consumption and is on a steady increase; therefore, it is required to prepare measures to increase the efficiency of energy consumption. One of the measures can be FEMS, which has functions such as real-time monitoring of energy use, energy consumption analysis, and control of various facilities, is attracting attention. Prediction of energy consumption among the various functions is an important first step in properly designing energy supplies and factory operations. In this paper, a study on energy consumption prediction was conducted for the purpose of application to FEMS. Machine learning technique was used to obtain predicted values by training the predictive model using available data, and LSTM algorithm was applied. To select a model suitable for the target food factory, the validity of the prediction model according to the combination of hyper parameters and variables was analyzed with CvRMSE and R 2 . By applying the optimal combination, the average error rate and accuracy of the prediction were confirmed through comparison between the predicted and measured values