Abstract: Heating prediction is essential for energy saving. However, it is difficult to some extent because the central heating system (CHS) is nonlinear with large-scaled delay and other characteristics. Based on physical laws of the CHS, a novel model combining pruning algorithm with the least mean square is proposed to solve the problem of delay and predict the temperature. The model with the delay taking into consideration conforms to the law of thermal transmission. The pruning algorithm lowers the computational complexity. Simulations using gradient descent least squared regression (LSR), support vector machine (SVM), least-squared support vector machine (LS-SVM) and BP neural network are put forward to evaluate the model and the results show that the model has low average relative error and mean squared error; and the model is simple and easy to calculate.

Key words: central heating system, prediction of secondary network, least mean square, pruning algorithm