Globalusis rostverkinių pamatų optimizavimas hibridiniu genetiniu algoritmu
| Author | Affiliation | |
|---|---|---|
Matematikos ir informatikos institutas | ||
| Year | Issue | Start Page | End Page |
|---|---|---|---|
2009 | 1(2) | 80 | 88 |
| URI | Access Rights |
|---|---|
| https://hdl.handle.net/20.500.14911/124393 | |
| Straipsnis | Viso teksto dokumentas (atviroji prieiga) / Full Text Document (Open Access) |
Straipsnio tikslas – sukurti technologiją polių padėtims rostverkiniuose pamatuose optimizuoti. Siekiama minimizuoti polių skaičių rostverke. Tikslo funkcija imama didžiausia atraminė poliaus reakcija, kuri optimizavimo metu yra minimizuojama. Iš pradžių uždavinys sprendžiamas genetiniu algoritmu, o kai toliau genetinį algoritmą naudoti tampa neefektyvu, gautas rezultatas gerinamas specifiniu atsitiktinės lokalios paieškos algoritmu. Analizuodami eksperimentinius duomenis autoriai bando nustatyti momentą, kada reikia pereiti nuo genetinio algoritmo naudojimo prie lokalios paieškos algoritmo, kad geresnis rezultatas būtų pasiekiamas per kuo trumpesnį laiką. Bendrą algoritmą, sujungiantį genetinį algoritmą su lokalios paieškos algoritmu, autoriai vadina hibridiniu genetiniu algoritmu.
The results of solution are the required number of piles and their positions.The entire optimization problem is solved in two steps. First, the grillage is transformed to a one-dimensional construct, and the optimizer decides about a routine solution (i.e. the positions of piles in this construct). Second, the backward transformation returns the pile positions into the two-dimensional grillage, and the “black-box” finite element program returns the corresponding objective function value. On the basis of this value the optimizer predicts the new positions of piles, etc. The finite element program idealizes the connecting beams as the beam elements and the piles – as the finite element mesh nodes with a given boundary conditions in form of vertical and rotational stiffnesses. The optimizing program is an elitist genetic algorithm or a random local search algorithm. At the beginning of problem solution the genetic algorithm is employed. In the optimization problems under consideration, the genetic algorithms usually demonstrate very fast convergence at the beginning of solution and slow non-monotonic convergence to a certain local solution point after some number of generations. When the further solution with a genetic algorithm refuses to improve the achieved answer, i.e. a certain local solution is obtained; the specific random search algorithm is used. The moment, at which the transition from genetic algorithm to the local search is optimal, is sought in the paper analyzing the experimental data. the hybrid genetic algorithm that combines the genetic algorithm itself and the local search is suggested for the optimization of grillages.