应用BP神经网络预测超临界CO_2的摩尔体积

Using BP Neural Network to Predict the Molar Volume of SCF-CO2

用BP神经网络,在1.017≤Pr≤8.134和1.019≤Tr≤5.917的范围内,对798组超临界CO2的P-V-T数据进行训练和预测,预测160组,平均相对误差为2.01％;用RKRKS及PR状态方程法计算这160组数据,平均相对误差分别为2.66％、3.24％和2.48％.表明神经网络法优于状态方程法。

BP Neural Network was used to predict the molar volume of supercritical Carbon Dioxide. In the range of 1.017≤pr≤8.134 and 1.019≤Tr≤5.917,798 experiment points were trained and predicted. The average relative error of predicting is 2.01%. Then RK EOS,RKS EOS and PR EOS are used to calculate these data,with average relative deviations being 2.66% ,3.24% and 2. 48% respectively. This method is proved to be superior to EOS.