Chemical process controller design using genetic programming
DP Searson, G.A. Montague, MJ Willis · 1998
In this article, we show that genetic programming (GP) can be used to design discrete-time dynamic controllers that offer similar performance to standard Proportional Integral Derivative (PID) controllers for a specific class of control objectives. Two processes are used as examples: an Auto-Regressive eXogeneous (ARX) system and a simulated non-linear Continuous Stirred Tank Reactor (CSTR). Additionally, some of the generalisation properties of the evolved controllers are examined and discussed. 1.