基于BP神经网络的R404A物性参数计算

Calculation of Thermal Properties of R404A Based on BP Neural Network

建立一个三层BP神经网络模型,以制冷剂R404A为研究对象,对其饱和液态的热物性参数:压力、密度、焓、熵进行模拟计算。将神经网络的计算结果与Refprop软件的结果进行对比后发现,压力、密度、焓、熵的平均计算误差为:5.9%,0.732%,1.74%,1.23%。证明该BP神经网络模型具有较高的精度。

Building a three-layer BP neural network model to calculate the thermal properties of R404 A,including pressure,density,enthalpy and entropy.The calculation is carried out under the saturation state of liquid.Compared the results of Refprop,it shows that the mean errors of difference respectively are 5.9%in pressure,0.732%in density,1.74%in enthalpy and 1.23%in entropy.Thus,the model can be used to calculate the thermal properties with high accuracy.