中文摘要：

为提高提纯塔出口二氧化碳纯度，根据实际生产数据运用人工神经网络方法建立了出口二氧化碳纯度与进塔纯度、进塔温度、塔顶压力、塔顶温度、塔釜加热温度、塔釜压力、塔釜温度7个因索之间的非线性模型．建立的模型能够有效描述出口二氧化碳纯度与各因素之间的关系，同时通过训练好的网络能够找到生产较高纯度二氧化碳的最佳控制点．与传统线性回归模型、对数回归模型相比，采用人工神经网络方法建立的模型非线性处理能力强，鲁棒性好，拟合精度高，计算速度快，预测和控制能力强．

英文摘要：

To improve the purity of carbon dioxide from the purifying column, a nonlinear model was established for describing the relationship between the carbon dioxide purity and the seven factors of feed purity, feed temperature, column top pressure, column top temperature, column reactor heating temperature, column reactor pressure and column reactor temperature built on the actual production data by using artificial neural network. The model can describe effectively the relationship between the carbon dioxide purity and other factors, and the best control point for the carbon dioxide production can be found by means of the trained network. The model based on the artificial neural network has advantages of robustness （the fitting is initial-value-free）, better fitting precision, strong ability to treat nonlinear model, fast computing, powerful prediction and strong ability to control compared with the conventional linear regression model and the logarithmic regression model.