非线性部分最小二乘法在催化剂活性建模中的应用

An Application of Nonlinear Partial Least Square in Catalyst Activity Modeling

结合统计回归与神经网络的优点，使用基于神经网络的非线性部分最小二乘回归法，建立了醋酸乙烯生产装置催化剂活性的非参数模型，模型的精度高且计算量较小。实际应用证明了方法的有效性。

A non parametric model of catalyst activity in the device of vinylacetate has been developed using nonlinear partial least square regression based on neural networks. Because of combining the merits of statistical regression and neural networks, the model has high accuracy and generosity. Practical application also proved its effectiveness.