A hybrid RF-LSTM based on CEEMDAN for improving the accuracy of building energy consumption prediction

An accurate method for building energy consumption prediction is important for building energy management systems. However, building energy consumption data often exhibits nonlinear and nonstationary patterns, which makes prediction more difficult. This study proposes a hybrid method of Random Forest (RF) and Long Short-Term Memory (LSTM) based on Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) to predict building energy consumption. In the first stage of our proposed method, the original energy consumption data is transformed into several components using CEEMDAN. Then, RF is used to predict the component with the highest frequency, and the remaining components are predicted using LSTM. In the last stage, the prediction results of all components are combined to obtain the final prediction results. The proposed method has been tested using real-world building energy consumption data. The experimental results demonstrate that the proposed method achieves better performance than the benchmark methods used for comparison.