One of the most important issues in the development of commercial transport aircraft is the reduction of the structural weight. To achieve this aim, both the structural design configuration as well as the selection of materials have to be considered in the structural optimisation process in addition to geometrical data. This problem becomes even more complex with composite structures, where laminate stacking sequences have to be taken into account. Thus, the design problem involves a combination of continuous and discrete variables. Evolutionary algorithms are well suited to handle such kind of problems, but with the drawback of needing to evaluate a large number of design configurations. That causes an enormous computational effort, particularly, if finite element analyses are involved. In this paper several methods to enhance the efficiency of such algorithms are presented. In detail these are a combined application of different types of evolutionary operators, an adaptation of the optimisation parameters itself and an approximate evaluation of design configurations during intermediate generations. The implemented approximation is based on a support vector machine classification approach. By the use of these methods the reliability of evolutionary algorithms to get the optimal solution in a single run is increased considerably. Furthermore, the number of exact evaluated design configurations is reduced by 60 %.
Optimisation of composite aircraft panels using efficiency enhancing evolutionary algorithms
Optimierung von Flugzeugbauteilen in Verbundbauweise mit einem effektivitätssteigernden evolutionären Algorithmus.
2006
10 Seiten, 7 Bilder, 1 Tabelle, 12 Quellen
Aufsatz (Konferenz)
Englisch
OPTIMIZATION OF COMPOSITE AIRCRAFT PANELS USING EFFICIENCY ENHANCING EVOLUTIONARY ALGORITHMS
British Library Conference Proceedings | 2006
|Optimization of Composite Aircraft Panels Using Evolutionary Computation Methods
British Library Conference Proceedings | 2000
|Optimisation of Composite Structures for Aeroelastic Applications Using Evolutionary Algorithms
British Library Conference Proceedings | 2008
|SURROGATE MODELLING FOR EFFICIENT DESIGN OPTIMISATION OF COMPOSITE AIRCRAFT FUSELAGE PANELS
British Library Conference Proceedings | 2010
|