Prediction of Building Energy Consumption Based on LSTM-SVR-Random Forest Hybrid Model

Accurate building energy prediction can help effectively allocate and utilize energy resources, reducing waste. The dataset used in this study contains data from 539 buildings and includes 12 variables. This study proposes a hybrid model of Support Vector Regression (SVR) and Long Short-Term Memory (LSTM) based on random forest optimization. The model combines the accuracy of SVR with the time series modeling ability of LSTM, which significantly improves the prediction accuracy of building energy consumption. Through an in-depth analysis of historical energy consumption data, this paper identifies key factors influencing energy use and employs the random forest algorithm for feature selection and model optimization. The experimental results show that the accuracy of the LSTM single model is approximately 54%, while the accuracy of the SVR single model is 15%. The hybrid model achieves an accuracy rate of more than 90% on both the training set and the test set, which shows that the SVR-LSTM hybrid model using Random Forest can effectively improve the prediction performance. This research provides important theoretical foundations and practical guidance for energy-efficient building design and smart energy management, while also offering data support for future policy-making, demonstrating broad application prospects.