ISSN:
1572-8145
Keywords:
Genetic algorithms
;
production systems
;
job scheduling
Source:
Springer Online Journal Archives 1860-2000
Topics:
Mechanical Engineering, Materials Science, Production Engineering, Mining and Metallurgy, Traffic Engineering, Precision Mechanics
Notes:
Abstract The paper discusses an evolutionary knowledge approach to intelligent problem solving. A rule-based production system is used to model the problem and the means by which the problem space should be searched. Search heuristics are modelled as production rules. These rules are redundant as there may be more than one view on the best method for building solutions. Some rules may have complex reasoning for their actions, others have none. Deciding which rule is most appropriate is solved by a genetic algorithm and ultimately only the ‘fitter’ rules will survive. The approach eliminates the necessity of designing problem specific search or variation operators, leaving the genetic algorithm to process patterns independent of the problem at hand. Learning methods and how they aid evolution is also discussed: they are Lamarckian learning and the Baldwin effect. The approach is tested on a scheduling problem.
Type of Medium:
Electronic Resource
URL:
http://dx.doi.org/10.1023/A:1008928804949
Permalink