Long-term prediction of building energy consumption based on improved LSTM and informer models

In recent years, with the rapid development of the global construction industry, the demand for energy has increased significantly, and building energy consumption has been rising. During the operational phase of buildings, there are various variable factors affecting long-term energy consumption. This paper introduces improved LSTM and Informer prediction models. On the one hand, the improved models integrate the Informer into the LSTM to complement each other, mining the potential correlations between features and capturing long-term time series characteristics. On the other hand, the Bayesian algorithm is used for hyperparameter tuning to avoid errors caused by manual parameter adjustments.