A bell-curve based algorithm for mixed continuous and discrete structural optimization

An evolutionary based strategy utilizing two normal distributions to generate children is developed to solve mixed integer nonlinear programming problems. This Bell-Curve Based (BCB) evolutionary algorithm is similar in spirit to (μ+μ) evolutionary strategies and evolutionary programs but with fewer...

Full description

Saved in:
Bibliographic Details
Published in:Structural and multidisciplinary optimization 2002-09, Vol.24 (2), p.98-105
Main Authors: KINCAID, R. K, WEBER, M, SOBIESZCZANSKI-SOBIESKI, J
Format: Article
Language:English
Subjects:
Actuators
Evolutionary algorithms
Exact sciences and technology
Fundamental areas of phenomenology (including applications)
Minimum weight
Minimum weight design
Mixed integer
Nonlinear programming
Optimization
Physics
Solid mechanics
Static elasticity
Static elasticity (thermoelasticity...)
Structural and continuum mechanics
Tabu search
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:An evolutionary based strategy utilizing two normal distributions to generate children is developed to solve mixed integer nonlinear programming problems. This Bell-Curve Based (BCB) evolutionary algorithm is similar in spirit to (μ+μ) evolutionary strategies and evolutionary programs but with fewer parameters to adjust and no mechanism for self adaptation. First, a new version of BCB to solve purely discrete optimization problems is described and its performance tested against a tabu search code for an actuator placement problem. Next, the performance of a combined version of discrete and continuous BCB is tested on two-dimensional shape problems and on a minimum weight hub design problem. In the latter case the discrete portion is the choice of the underlying beam shape (I, triangular, circular, rectangular, or U).
ISSN:1615-147X
1615-1488
DOI:10.1007/s00158-002-0220-4