针对除霜控制方法的自适应性需求，本文提出一种制热能力衰减(Degradation of Heating Capacity，DHC)方法识别结霜状态，并基于全连接神经网络(Fully Connected Neural Network，FNN)分类模型开展除霜效果水平预测研究。结果表明:在空气源热泵的监测案例中，所提出的DHC方法能有效识别结霜状态，且在测试集中FNN分类模型对除霜效果水平识别的准确率达到91。43％。与原始除霜控制方法相比，整个供暖季的除霜频率、热量损失和功耗损失分别降低66。3％、1775 MJ和1829 MJ，同时季节性能参数SCOP提高8。6％。研究结果为ASHP系统的除霜控制方法在实际运行中的实施与优化提供了一条有效途径。
A Data-driven Evaluating Method on the Defrosting Effect of the Air Source Heat Pump System
Based on the adaptive demand of the defrosting control method,the degradation of heating capacity(DHC)method was developed in this work to identify the frosty state,and the defrosting effect was evaluated adopting a fully connected neural network(FNN)classification model.Results indicated that in the monitoring case of the ASHP system,the proposed DHC method can effectively identify the frosty state,and the defrosting effect recognition accuracy achieved 91.3％for the trained FNN classification model in the testing data set.Compared with the original defrosting control method,the defrosting frequency,heating loss and power consumption were respectively reduced by 66.3％,1775 MJ and 1829 MJ,and the SCOP was increased by 8.6％throughout the heating season.The promising results in this work will provide an innovative approach for the implementation and optimization of the defrosting control strategy of the ASHP system in practical operation.
air source heat pumpdefrostingdata-drivendegradation of heating capacity
NETL
NSTL
万方数据
