基于混合模型的中央空调冷源系统运行能耗优化

Optimization of Central Air Conditioning Cold Source System based on Hybrid Model

为优化中央空调冷源系统运行能耗,本文分别建立了中央空调冷源系统运行能耗预测灰箱模型和BP神经网络模型,对比分析了灰箱模型与BP神经网络模型的能耗预测性能,并基于K-means聚类算法提出了将灰箱模型和BP神经网络模型相结合的能耗预测混合模型。结合中央空调冷源系统能耗预测混合模型,以模型可控输入变量为优化变量,对中央空调冷源系统进行节能优化。结果表明:对比单独使用灰箱模型或BP神经网络模型,中央空调冷源系统混合模型能耗预测精度提升了27.7%和33.85%。对比冷源系统优化前能耗,优化后的中央空调冷源系统运行能耗平均降低了8.2%。

For optimizing the energy consumption of central air-conditioning cold source system, this paper establishes grey box model and BP neural network model to predict system energy consumption and compares the prediction accuracy of two models. A hybrid model is proposed to combine gray box model and BP neural network model based on K-means cluster algorithm. With the energy consumption prediction hybrid model, the paper optimizes the cold source system operation which controllable variables of the model input are used as optimization variables. The results show that compared with gray box model and BP neural network model alone, hybrid model prediction accuracy of cold source system energy consumption increases 27.7% and 33.85% on average. Compared with the cold source system energy consumption before optimization, the energy consumption after optimization decreases 8.2% on average.