Genetic algorithms and their potential for use in process control : A case study
To exploit genetic algorithms (GA) for process control, it is necessary to find a way to encode, in an efficient and generic manner, all relevant control parameters into one or more individuals and to define a fitness function, based on the encoding schema, which measures the quality of control performed. Three methods have been investigated: - Pure GA: the "classical application" of GA to process control; - Hybrid GA: the combined use of a conventional controller and a GA-based system; - the GA-based tesselation of the state space. These three approaches and the results of their application to the Tank case study are described in this paper. The results show that, even with certain simplistic assumptions, some of the investigated approaches achieve satisfactory control while fulfilling the two requirements of "on-line performance" and "no a priori knowledge", without the need for a training stage. In particular, the GA-based tesselation approach appears very promising.
Bibliographic Reference: Paper presented: 4th International Conference on Genetic Algorithms, San Diego, California (US), July 13-16, 1991
Availability: Available from (1) as Paper EN 35985 ORA
Record Number: 199110419 / Last updated on: 1994-12-02
Original language: en
Available languages: en