石油馏分平均相对分子质量预测的神经网络方法

Prediction of the Average Relative Molecular Mass of Petroleum Fractions by Artificial Nerve Network

将误差反向传播（ＥＢＰ）人工神经网络应用于石油馏分平均相对分子质量的预测上．对在７３～４８０℃沸点温度范围内１９组石油馏分平均相对分子质量和密度均值的研究表明，神经网络方法性能优越，具有很强的推广预测能力，可望成为石油馏分平均相对分子质量预测的有效手段．

Abstract:The artificial nerve network with error backpropagation is used for the prediction of average relative molecular mass of petroleum fractions. The predicted results of 19 groups of petroleum fractions whose boiling temperatures are in 73-480℃ show that the nerve network is effective to the prediction of the average relative molecular mass and average density of petroleum fractions.