A direct-expansion solar-assisted heat pump (DX-SAHP) system using R290 (propane) is studied experimentally. A Coriolis mass flow meter is employed to measure the R290 mass flow rate, and the artificial neural network (ANN) model is used to predict it. Based on numerous experiments, the 5 independent variables are chosen as the input parameters of the ANN model, including the ambient temperature, solar radiation intensity, electronic expansion valve (EEV) opening, compressor frequency and water temperature. The appropriate number of the hidden layer neurons is obtained. The results show that the Mean Relative Error (MRE), Root Mean Square Error (RMSE), and Standard Deviation (SD) of the ANN model are 0.0012, 0.4139 kg h⁻¹ and 0.0447, respectively. Above 97% of prediction results agree well with experimental data within a maximum error of 10%. Furthermore, as the ambient temperature increases, the refrigerant mass flow rate increases. As the solar radiation intensity increases, the refrigerant mass flow rate increases on the whole, and the effect of solar radiation intensity is weakened with the increment of ambient temperature. At a higher level of compressor frequency, the variation in the refrigerant mass flow rate is approximately linear with the ambient temperature, solar radiation intensity, EEV opening or water temperature. With the proposed ANN prediction model, the refrigerant mass flow rate of the DX-SAHP system can be quickly got, which will be very useful in system efficient operation.

实验研究了使用R290（丙烷）的直接膨胀式太阳能辅助热泵（DX-SAHP）系统。使用科里奥利质量流量计来测量R290质量流量，并使用人工神经网络（ANN）模型对其进行预测。基于大量实验，选择了5个独立变量作为ANN模型的输入参数，包括环境温度，太阳辐射强度，电子膨胀阀（EEV）开度，压缩机频率和水温。获得适当数量的隐藏层神经元。结果显示，ANN模型的平均相对误差（MRE），均方根误差（RMSE）和标准偏差（SD）为0.0012、0.4139 kg h ^-1和0.0447。超过97％的预测结果与实验数据非常吻合，最大误差为10％。此外，随着环境温度升高，制冷剂质量流量增加。随着太阳辐射强度的增加，制冷剂质量流量总体上增加，并且太阳辐射强度的影响随着环境温度的升高而减弱。在较高的压缩机频率水平下，制冷剂质量流量的变化与环境温度，太阳辐射强度，EEV开度或水温大致成线性关系。利用所提出的ANN预测模型，可以快速获得DX-SAHP系统的制冷剂质量流量，这对于系统高效运行非常有用。