建筑空间负荷预测方法
Building load prediction methods
摘要：
对建筑空间负荷计算方法的研究现状进行了分析总结，归纳出3类主要的负荷计算方法：基于机理建模的计算方法，基于数据学习的逐时预测方法，基于软测量的实时预测方法。基于机理建模的计算方法比较成熟，准确性高，比较适合在建筑空调系统的设计阶段进行模拟分析。后2种方法适用于空调运行阶段的实时调控，具有较大的发展潜力，但基于数据学习的逐时预测方法需要进一步加强模型泛化能力及动态实时性方面的研究，基于软测量的实时预测方法则需要在模型准确性、参数辨识有效性等方向重点关注。
Abstract：
Reviews the state-of-the-art of building load calculation methods, and summarizes these methods into three categories, including the calculation method on physical model, the hourly prediction method on data learning and the real-time prediction method on soft sensing. The calculation method on physical model has been well developed with high accuracy, which is suitable for simulation analysis in the design stage of air conditioning system. The other two categories of methods are suitable for real-time control in the running stage with great practical potential. But the hourly prediction method on data learning needs to solve some problems such as model generalization, dynamic and real-time ability, and the real-time prediction method on soft sensing needs to pay more attention to model accuracy and to develop more effective parameter identification techniques.
Keywords:building, central air conditioning, energy conservation, cooling/heating load, prediction, soft sensing