Prace doktorskie / Habilitacyjne (WTMiT)

Stały URI dla kolekcji

https://oa.zut.edu.pl/handle/20.500.12539/204

Przeglądaj

Przeglądaj Prace doktorskie / Habilitacyjne (WTMiT) wg Temat "optymalizacja wielokryterialna"

Teraz wyświetlane 1 - 1 z 1
  • Miniatura
    PozycjaOpen Access
    Wybrane problemy optymalizacji wielokryterialnej we wstępnym projektowaniu konstrukcji kadłuba statków morskich
    (Wydawnictwo Uczelniane Zachodniopomorskiego Uniwersytetu Technologicznego w Szczecinie, 2012) Sekulski, Zbigniew; Zachodniopomorski Uniwersytet Technologiczny w Szczecinie. Wydział Techniki Morskiej i Transportu
    In the continuously changing economic situation the methods for ship structural design used nowadays are sometimes insufficient. A need for modernization of the cognitive apparatus is more and more apparent in the ship structural design to discover and conceptualize various aspects of designing. New methods can be attempted to be found or change the ones used so far. It is possible and may turn efficient application of methods developed in other fields. The results of the investigations prove that in the ship design and, especially, in the ship structural design, multi-objective evolutionary algorithms may be especially useful. Applications of these algorithms, even not many numerous, have been attempted, also by the author, since the 1990th confirmed their effectiveness also in the ship structural design. In this context, an objective of the dissertation is to develop a method for evolutionary multi objective optimization of complex structures such as ship hull structures. Multi-objective optimization of complex objects such as the hulls of seagoing vessels, despite more and more greater research and computational potential still encounters many barriers making practical application difficult even at the preliminary design where the models are significantly simpler than the models at the next design stages. The difficulties make investigations and applications related to the ship hull optimization focus on single-objective despite the fact that the most real problems in the ship structural design call for considering more than one optimization objective. From the point of view of the problems related to the ship hull structural design searching optimal solutions is performed in the following four areas: shape optimization, material optimization, topology optimization as well as optimization of dimensions of struc- tural elements. Due to the specific features of the optimization problems solved in each of these areas it is thus impossible to formulate and solve a single optimization problem covering all the areas. A common practice is therefore that in each of the areas the optimization problem is formulated and solved independently of the others. The division is disadvantageous, however, for the correct designing and the reason can be the lack of suitable tools for simultaneous designing-optimization in all areas. Formulation of the separate optimization problems makes the ship design both longer and more expensive as well as delimits or disables effective search for satisfactory optimal solutions. Yet there is no promising alternative for such a procedure, therefore it seems advisable to attempt to solve unified shape-material- topology-dimension optimization problem. So far the effective tools have not been developed. Investigation towards developing methods for multi-objective optimization seems jus- tified. Due to the complexity of the problem it is reduced to topology-shape optimization treating it as the first step on the way towards formulation and solution of the full unified shape-material-topology-dimension optimization problem. In the context of the general theory of evolutionary algorithms of multi-objective optimization an original method was developed and applied based on the genetic algorithm. In the method evolution is controlled by the scalar fitting function allowing for consideration of the evaluation objective, level of violating the formulated constraints, internal structure of the set of permissible solutions as well as distances of the generated solutions from the as- ymptotic solution defined in the dissertation. A set of non-dominated solution in the Pareto sense was taken as the result of solution of the multi-objective problem of the multi- objective optimization. The set is interpreted as an approximation of (unknown) real set of Pareto-optimal solutions. A number of original computational procedures were developed which were applied in the computer codes. Problems of assessment of solution quality and efficiency of the multi-objective evolutionary algorithms, controversial but rarely referred to in the literature, were discussed in details. Quality measures were suggested which allow for efficiency assessment of the developed algorithm. To verify the computer code a computational model of the fast passenger-car ferry hull structure was developed. A number of computer simulations were performed using the model. The results, which were found the most important, were presented and discussed in details. Application of the method to optimization of the twin-hull high-speed vessel allowed to demonstrate the specific features of the multi-objective evolutionary optimization methods. It was also possible to formulate the general conclusions and assess the applicability of the method to design analyses of such structures. The method of multi-objective evolutionary optimization, presented in the dissertation, can be, according to the author’s opinion, recommended to be applied to the ship structural preliminary design. It can also be applied to optimization of the other structures after implementation of computational procedures specific for the solved problem.