Short and medium-term prediction of heating load for office buildings in Tianjin

Analyses the relationship between the average daily dry-bulb temperature and holiday effect and the daily average load, and establishes the multivariate nonlinear regression model with residual error correction to predict the daily average load. Selects date and day type as the discriminatory factors of “similar day”, and predicts the hourly heating load according to load distribution coefficient of the similar day. Case studies show that the prediction model has high prediction accuracy for the office buildings.