A Research on Building Energy Consumption Prediction Method Based on GBRT-LSTM

Building energy consumption prediction plays an important role in improving the scheduling efficiency of the power grid, but the building energy consumption data pattern is very complex and it is difficult to establish a high-precision prediction model. To address the above problems, this study proposes a building energy consumption prediction method combining gradient boosting regression tree and long and short-term memory network. Firstly, adaptive noise-complete ensemble empirical modal decomposition is used to decompose the energy consumption data into multiple components. For the component with the highest frequency, a gradient boosted regression tree was used to predict it. For the remaining components, long and short-term memory networks were used to predict them. Finally, the results of the gradient boosting regression tree and the long and short-term memory network predictions were summed to obtain the prediction of the final energy consumption. In order to validate the effectiveness of the proposed method in this study, several real-world energy consumption data were used to test the proposed method and compare it with a variety of prominent machine learning methods. The experimental results show that the proposed method performs better than the compared methods.