Comparison of several centralized heating load forecasting models

 In order to improve the accuracy of the heating load prediction, adds indoor temperature factor into the prediction models. In addition, applies multivariable linear regression (MLR), BP neural network and support vector machine regression based on grid search (GS-SVR) to the heating load prediction in the next seven days. The obtained results show that GS-SVR model prediction is more accurate than MLR and BP neural network, which can be adopted to guide the engineering practice.