Use this URL to cite or link to this record in EThOS:
Title: New genetic algorithms for constrained optimisation and applications to design of composite laminates
Author: Yu, Tat Loong
ISNI:       0000 0004 2746 2715
Awarding Body: Cardiff University
Current Institution: Cardiff University
Date of Award: 2004
Availability of Full Text:
Access from EThOS:
Access from Institution:
A general purpose constraint handling technique for genetic algorithms (GA) is developed by borrowing principles from multi-objective optimisation. This is in view of the many issues still facing constraint handling in GA, particularly in the number of control parameters that overwhelms the user, as well as other GA parameters, which are currently lacking in heuristics to guide successful implementations. Constraints may be handled as individual objectives of decreasing priorities or by a weighted-sum measurement of normalised violation, as would be done in multi-objective scenarios, with full consideration of the main cost function. Rather than the unnecessary specialisation seen in many new heuristics proposed for GA, the simplicity, generality and flexibility of the technique is maintained, where several options such as partial or full constraint evaluation, tangible or Pareto-ranked fitness, and implicit dominance evaluation are presented. By reducing the number of constraint evaluations, these options increase the probability of discovering optimal regions, and hence increase GA efficiency. Studies in applications to a constrained numerical problem, and to the design of realistic composite laminate plates and structures, serve to demonstrate the ease of implementation and general reliability in heavily constrained problems. The difference in the dynamics of partial or full violation knowledge showed that while the former reduced the overall number of constraint evaluations performed, the latter compromises for the expense of full constraint evaluations in the reduced number of GA generations, whether in terms of discovering feasible regions or optimal solutions. The benefit of partial or full constraint evaluations is subjective, as it ultimately depends on the trade-off in the computational cost of constraint evaluations and GA search.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available