Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 82
Edited by: B.H.V. Topping
Paper 22

A new Parallel Genetic Algorithm Scheme in Structural Design Optimisation

T. Talaslioglu

Vocational Training College, Cukurova University, Osmaniye, Turkey

Full Bibliographic Reference for this paper
T. Talaslioglu, "A new Parallel Genetic Algorithm Scheme in Structural Design Optimisation", in B.H.V. Topping, (Editor), "Proceedings of the Eighth International Conference on the Application of Artificial Intelligence to Civil, Structural and Environmental Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 22, 2005. doi:10.4203/ccp.82.22
Keywords: genetic algorithm, optimisation, truss, frame, semi-rigid frame.

In the construction of multi-story buildings on confined sites and of industrial type buildings in open land, the rolled steel profiles are preferred for strength-related, economic and aesthetic reasons. The choice of in cross-sectional dimensions and strength of the rolled steel profiles reflects itself, mathematically, as an optimisation problem. The complexity of the structural optimisation problem is determined by the mesh attributes (number and connectivity of members) and problem attributes (the conditions of constraints whether the design variable are continuous or discrete). The factors taken into consideration in the choice of optimisation method are: i) mathematical complexity, ii) ability to handle all variable types, iii) computational cost, iv) ability to convert the optimisation problem from constrained case to unconstrained, v) ease of computer coding. GAs can constitute an excellent choice in view of the above-mentioned considerations. Different types of GAs are successfully used in structural design optimisation problems.

GAs are capable of finding promising sub-regions in the complex search region through a randomised migration performed among the individuals of the population. However, once a state of equilibrium is reached in a certain sub-region, the search process stagnates. This situation is called "immature convergence". Discovering the correct trajectory to the global minima is possible only through a communication among populations. If the variations among the populations and the diversity among individuals are kept, the search for the global minima, will be more effective. In addition, these factors also help to elevate the exploitation during the search. But in traditional GAs, each additional trials is likely to lead already observed sub-regions of the topography. This causes a stagnation in the migration. Previous, attempts have been made to restrict either the selection procedure (crowding models) or the mating procedure (assorted mating, local mating) as a remedy for immature convergence [1]. Also, an approach which the population is explicitly divided into sub-populations is investigated as a method [2,3]. Most recent works related with GAs is almost entirely devoted to parallel systems. Parallel systems preserve diversity and ensure perpetual novelty, thereby disseminating the different characteristics of the individuals to the entire populations.

This paper presents a new parallel genetic algorithm (PGA) scheme (AMGAwFBD/I). In this scheme, the movement in the search region is performed through an asynchronous migration having an assimilation property in two distinct parallel phases. An asynchronous migration based on feasible solutions obtained from the search, controls both infusion and diffusion processes using a control mechanism. In addition, use of traditional genetic algorithm operators is included in the search. The AMGAwFBD/I scheme is evaluated on five different structural design optimisation problems with design variables of a discrete type. In all example problems this method produced improved results compared to previous work. The AMGAwFBD/I method is demonstrated to be a robust PGA method.

Smith R.E., Forrest S., Perelson A.S., "Searching for diverse, co-operative populations with genetic algorithms", Evolutionary Computation, 1993; 1(2): 127-149. doi:10.1162/evco.1993.1.2.127
Leite, J.P.B., Topping, B.H.V., "Improved Genetic Operators for Structural Optimisation", Advances in Engineering Software, 1998, 29 (7-9): 529-562. doi:10.1016/S0965-9978(98)00021-0
Topping, B.H.V., Leite, J.P.B., "Parallel Genetic models for Structural Optimisation", Int. J. of Engineering Optimisation, 1998, 31(1): 65-99. doi:10.1080/03052159808941366

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £80 +P&P)