基于BP神经网络的冷水机组能效预测方法

The COP Prediction Method of Chillers Based on the BP Neural Network

针对冷水机组能效物理模型及辨识模型结构复杂,模型参数难以获取等特点,提出了基于BP神经网络的冷水机组能效预测方法,并以某酒店离心式冷水机组为例,建立了其冷水机组能效预测模型,并利用实际运行数据对模型进行了训练及验证,同时介绍了该模型神经网络结构参数的设计及冷水机组能效预测方法实现过程,线性回归分析结果表明:提出的能效预测方法能够准确地预测冷水机组的能效,方法简单、实用,具有一定的适用性。

The physical and identification model of COP for chillers are always described by complex formulation, and require a great number of parameters to measure or identify, so it is not easy to fulfill this study. A prediction method of COP for a real chiller plant has been proposed in this paper, and also be trained and tested using the operation data of the chiller plant. In the meantime, the design process of the structure parameter and realize process of the method have been introduced in this paper, the consequence of the analysis of train and test using linear regression technique shows that the method proposed in this paper can predict the COP of the chiller accurately and quickly, and the method is simple but it is useful to the COP prediction of chillers.