基于线性和非线性判别的工业过程建模方法

A Linear-nonlinear Discriminant Analysis Based on Statistical Modeling Method for Industrial Processes

在传统的工业过程建模方法的基础上,针对变量间往往存在复杂的数据特性的问题,提出一种线性和非线性关系的判别方法。以多元统计分析为基础,对变量之间的相关性进行分析,有效区分了线性变量和非线性变量。在此基础上提出了相应的建模方法,提高了模型精度。通过仿真验证,对过程进行线性和非线性判别有助于提高过程模型的精度。

Based on the traditional industrial process modeling methods,focusing on the complex data characteristic problem,a linear and nonlinear discriminant analysis algorithm is proposed.This method adopts multivariate statistical analysis to analyze variable correlations,which can effectively distinguish linear variables and nonlinear variables.Based on it,corresponding modeling strategy is proposed.The method is verified by simulation example and conclusion can be drawn that modeling accuracy can be improved.