Predicting energy demands based on the past energy consumption can allow a reasonable allocation of energy resource to avoid waste and improve utilization. To this end, linear or nonlinear forecasting models are applied. Some researchers use support vector regression models to deal with the energy consumption prediction problem as they can handle with nonlinear problems through their kernel function. However, using fuzzy rule-based models based on the granulation–degranulation mechanism to predict energy consumption can better deal with the nonlinear data and further improve the robustness and the accuracy of prediction compared with the support vector regression models. In this paper we apply a first-order fuzzy rule-based model to predict the energy data. Firstly, the data is granulated in the input space, and then the number of rules is determined according to the error value between the estimated value and the actual value. The prediction task can be completed based on a small amount of input data. It has good interpretability and delivers superior predictive performance. The experimental results show that the improvement of performance index MAE of the first-order fuzzy rule-based model is 18.59%, 37.58%, 25.82% and 8.43% better than that of the Lasso model, support vector regression, zero-order fuzzy rule-based model and LSTM-RNN model, respectively, on the testing data.