Título:
|
Global search methods for nonlinear optimisation: a new probabilistic-stochastic approach
|
Autor/a:
|
Bugeda Castelltort, Gabriel; Balsa-Canto, E; Thamotheram, C; Oñate Ibáñez de Navarra, Eugenio; Zárate Araiza, José Francisco
|
Otros autores:
|
Universitat Politècnica de Catalunya. Departament d'Enginyeria Civil i Ambiental; Universitat Politècnica de Catalunya. (MC)2 - Grup de Mecànica Computacional en Medis Continus; Universitat Politècnica de Catalunya. GMNE - Grup de Mètodes Numèrics en Enginyeria; Universitat Politècnica de Catalunya. LITEM - Laboratori per a la Innovació Tecnològica d'Estructures i Materials |
Abstract:
|
In this work the problem of overcoming local minima in the solution of nonlinear optimisation problems is addressed. As a first step, the existing nonlinear local and global optimisation methods are reviewed so as to identify their advantages and disadvantages. Then, the major capabilities of a number of successful methods such as genetic, deterministic global optimisation methods and simmulated annealing, are combined to develop an alternative global optimisation approach based on a Stochastic-Probabilistic heuristic.
The capabilities, in terms of robustness and efficiency, of this new approach are validated through the solution of a number of nonlinear optimisation problems. A well know evolutionary technique (Differential Evolution) is also considered for the solution of these case studies offering a better insight of the possibilities of the method proposed here. |
Materia(s):
|
-Àrees temàtiques de la UPC::Matemàtiques i estadística::Investigació operativa::Simulació -Nonconvex programming -global optimisation -stochastic-probabilistic approaches -Monte-Carlo -optimal design -shape optimisation -Optimització matemàtica |
Derechos:
|
|
Tipo de documento:
|
Artículo - Versión publicada Objeto de conferencia |
Editor:
|
International Centre for Numerical Methods in Engineering (CIMNE)
|
Compartir:
|
|