PCA-ANN方法在PTA平均粒径软测量中的应用

The application of PCA-ANN method to PTA average particle size soft sensor

在精对苯二甲酸(PTA)生产过程中,影响产品粒度的因素多且复杂。采用主元分析(PCA)方法对高维输入变量进行降维处理,利用PCA与神经网络相结合(PCA-ANN)方法建立了PTA产品平均粒径的软测量模型,模型误差在5%以内,工业应用结果验证了方法的有效性。

In the production process of PTA , there are many factors that have effect on the product particle size. Dimension decreased treatment was carried out on high dimensional input variable by using primary component analysis （PCA） method. The soft sensor model of PTA average particle size was established by PCA-ANN method. The model error is within 5%, The efficency of the method was tested and verified by industrial application results.