A Taxonomy for artificial embryogeny
Artificial Life
A fast and elitist multiobjective genetic algorithm: NSGA-II
IEEE Transactions on Evolutionary Computation
On learning to generate wind farm layouts
Proceedings of the 15th annual conference on Genetic and evolutionary computation
| Hi-index | 0.00 |
Polymer microstructured optical fibres are a relatively recent development in optical fibre technology, supporting a wide variety of microstructure fibre geometries, when compared to the more commonly used silica. In order to meet the automated design requirements for such complex fibres, a representation was developed which can describe radially symmetric microstructured fibres of different complexities; from simple hexagonal designs with very few holes, to large arrays of hundreds of holes. This representation uses an embryogeny, where the complex phenotype is 'grown' from a simpler genotype, and the resulting complexity is primarily a feature of the reuse of gene elements that describe the microstructure elements. Most importantly, the growth process results in the automatic satisfaction of manufacturing constraints. In conjunction with a multi-objective genetic algorithm, this formed a robust algorithm for the design of microstructured fibres for particular applications of interest. In this paper the algorithm is used to design one of the most common types of microstructured fibres - single-moded fibres. Various types of single-moded designs that have not been encountered in the literature were discovered, identifying new 'design themes'. One of the designs was subsequently manufactured, the details of which are included.