Research On Optimization Of CO₂ Heat Pump Integrated Heating System Based On Time Series Forecast

Heating,ventilation,and air-conditioning（HVAC）systems and hot water supply have become an increasingly important part of people’s lives and have consumed a large amount of energy.Improving the energy efficiency of the heating system has huge energy saving potential.In this thesis,a CO₂ heat pump integrated heating system located in central Norway is analyzed.By analyzing its operating data,the current problem in the system is found,a corresponding solution is proposed.As the key part of the solution,the ventilation system heat demand power forecast is fully illustrated in this thesis.The work of this thesis mainly draws the following two conclusions:（1）The models established based on MLP and LSTM outperform the benchmark model established by the persistence method.The ability of automatically feature learning and arbitrary mapping function learning is verified.The performance of the LSTM model outperforms the MLP model,which shows the advantage of LSTM networks in processing time series data.（2）Many factors contribute to the changing of the ventilation system heat demand,including human behavior,weather changes,the status of the energy storage system,etc.By utilizing multi-dimensional data to train the model,valuable information is provided to the model,which further improves the forecast result of the model.Finally,the multivariable LSTM model achieves the best result in predicting the ventilation system heat demand power on the validation set.The RMSE of the forecast result is 5.12 and the MAPE is 10.67%.The RMSE is 48.70%better compared to the benchmark model.