Bayesian Optimization-Based LSTM for Short-Term Heating Load Forecasting

With the increase in population and progress of industrialization, rational use energy heating systems has become a research topic for many scholars. The accurate prediction heat load provides us with scientific solution. Due to complexity difficulty forecasting systems, this paper proposes short-term method based on Bayesian algorithm-optimized long- memory network (BO-LSTM). moving average data smoothing is used eliminate noise from data. Pearson’s correlation analysis determine inputs model. Finally, outdoor temperature previous period are selected as root mean square error (RMSE) main evaluation index, absolute (MAE), bias (MBE), coefficient determination (R2) auxiliary indexes. It was found that RMSE asynchronous length model decreased, proving general practicability method. In conclusion, proposed simple universal.