GP - Genetic Programming for Model Generation and Model Predictive Control
Industrial sponsor: Mistubishi Chemical Corporation, Japan
Genetic
Programming (GP) is an evolutionary-computational method
relying
on tree-like building blocks, which manipulates populations of structures of
varying length and complexity.
It is therefore an appropriate approach for process modeling, in
that it permits for the parameters and structure of an empirical model to be
optimized.
Our
research has led to the development of an improved GP to facilitate the
generation of steady state nonlinear empirical models for process analysis
and optimization.
The
work has been extended to automatically
generate an empirical dynamic process model, and its use in a nonlinear
model predictive control (NMPC) strategy.
Since transient models are required for NMPC, the GP derives both a
model structure and its parameter values in such a way that the process
trajectory is predicted accurately.
The genetic programming approach has been applied on several
multivariable process examples such as the run-to-run control of a complete
process for photolithography, and temperature uniformity in rapid thermal
processing (RTP).
Recent research has focused on the application of genetic programming to the
automated design of nonlinear controllers based on a Lyapunov approach.
Potential industrial applications: Automated nonlinear modelling and rapid prototyping of model-based
control.
Selected References
S.
Lakshminarayanan, H. Fujii, B. Grosman, E. Dassau and D. R. Lewin, “New
Product Design via Analysis of Historical Databases”, Comput. Chem. Eng.,
24(2-7), 671-676 (2000).
[Download paper
320kb]
B.
Grosman and D.
R. Lewin, “Automated Nonlinear Model Predictive Control using Genetic
Programming,” Comput. Chem. Eng., 26(4-5), 631-640 (2002). [Download paper
292kb]
B. Grosman and D.
R. Lewin,
“Adaptive
Genetic Programming for Steady-state Process Modeling,”
Comput.
Chem. Eng.,
28(12), 2779-2790 (2004).
B. Grosman and D. R. Lewin, “Automatic Generation of Lyapunov Functions
using Genetic Programming,” to be presented the 16th IFAC World Congress,
Prague (2005).
[Download
paper 477kb]
Researcher:
Benny Grosman