Conclusions
By establishing sub-metering platform, real-time monitoring of building power consumption can be achieved. Massive power consumption data will be provided, laying a good foundation for the future research. By means of regression algorithm, the short-time prediction of power consumption in large-scale public buildings can also be realized, offering a gist for the building facility managers and relevant decision-makers. But we should make classification of the public buildings before the regression because of their variance in construction features and operating characteristics, therefore improving the accuracy of the prediction.
Short-term prediction models of building power consumption in cooling and heating season are established according to the sub-metering consumption of six large-scale public buildings in Shanghai. The weekly power consumption of office buildings and building complexes mainly depends on the factors of temperature and workdays in a week, while, for the commercial buildings, it only depends on the temperature for the impact of the workdays in whole week is not remarkable. The impact of temperature can be described by CDH and HDH. The prediction models are validated to have great accuracy and general applicability which provide a good foundation for the large-scale buildings in diagnosing, energy monitoring, energy-saving reform and regulatory work.
Acknowledgements
The authors gratefully acknowledge the support provided by Shanghai Science and Technology Committee through the Project 'Innovative Research and Practice of Energy Saving in Large-scale Public Building and Energy Monitoring Platform in Shanghai Central Area'.