Symbolic regression modeling of blown film process effects
Arthur K. Kordon, Ching-Tai Lue · 2004
The potential of symbolic regression for automatic generation of process effects empirical models has been explored on a real industrial case study. A methodology based on nonlinear variable selection and model derivation by genetic programming has been defined and successfully applied for blown film process effects modeling. The derived nonlinear models are simple, have better performance than the linear models, and predicted behavior in accordance with the process physics.