Heating load prediction of building complexes based on subspace identification method

The state space model is established according to the building thermodynamic model and the heating load calculation model.The heating load prediction model of building complexes is constructed by using the subspace identification method with unspecified and specified state variables,respectively.Taking a heat exchange station of a residential district in Harbin as an example,the heating load prediction results of the two methods and the effects of training strategies and prediction duration on the prediction accuracy are verified and compared.The results show that the mean absolute percentage error (MAPE) of the heating load prediction using the subspace identification method with specified state variables is reduced by 10.00% compared to that with unspecified state variables.The medium-term prediction effect of the subspace identification method with unspecified state variables is better than the short-term prediction effect,and the MAPE is reduced by 13.34%.