利用人工神经网络模型 ,根据催化裂化原料油的碳元素质量分数 (ωC)、氢元素质量分数 (ωH)、密度 (ρ2 04 )、运动粘度 (ν10 0 )四参数对催化裂化原料油的结构族组成进行预测 ,获得了较高的精度 ,计算值与试验值相比 ,CP、CN、CA 的平均绝对误差分别为 0 .36 %、0 .16 %、0 .15 %。用该模型对未参加训练的 4种油样进行检验 ,预测结果的平均相对误差为 4.0 5 % ,表明此方法具有较高的拟合精度和较好的预测能力 Based on the artificial neural network model and the four parameters of the feedstock oil, such as the carbon content (ωC), hydrogen content (ωH), density (ρ2 04) and kinematic viscosity (ν10 0) Compared with the experimental data, the average absolute errors of CP, CN and CA were 0.36%, 0.16% and 0.15%, respectively. The model was used to test the four oil samples that did not participate in the training. The average relative error of the prediction results was 4.05%, indicating that this method has higher fitting accuracy and better prediction ability