Genetic algorithms for optimization in predictive control

Genetic algorithms (GAs) are optimization methods inspired by natural biological evolution. GAs have been successfully applied to a variety of complex optimization problems where other techniques have often failed. The aim of this paper is to investigate the use of GAs for optimization in nonlinear model-based predictive control, Advanced genetic operators and other new features are introduced to increase the efficiency of the genetic search. In order to deal with real-time constraints, termination conditions are proposed to abort the evolution, once a defined level of optimality is reached. Simulated pressure dynamics of a batch fermenter are considered as an example of a highly nonlinear system. Simulation results with GAs are compared with the branch-and-bound method, in terms of the control accuracy and computational costs achieved. Copyright (C) 1997 Elsevier Science Ltd.