Binary Particle Swarm Optimization for variable selection in Partial Least Squares Regression with application to Production Quality Modeling

摘要

In order to build the regression model between production process parameters and quality results of products in cold rolling strip hot-dip galvanizing, particle swarm optimization algorithm is used to perform variable selection in partial least squares to determine which some variables were be chosen. The cold rolling strip hot-dip galvanizing data consists of nine components of active substances, optimization variables selection solutions are able to improve the regression model which can eliminate some unimportant or uninformative variables and obtain the more simple and optimal prediction model. Because the obtained model contains few variables, it is easy to analyse or explains the variables effect to the model. Hence the method can be used to discriminate or determine the production process parameters effectively.