Keywords: optimization, genetic algorithm, value encoding, discrete design variable, truss structure, coded program.

There are several techniques used in engineering optimization. Genetic algorithm is one of these techniques based on the mechanics of natural selection. Most of methods assume that the design variables are continuous. Due to the availability of components in standard sizes design variables are discrete [1]. Unlike other optimization techniques, genetic algorithms use discrete design variables. Genetic algorithm uses a population and a set of natural selection operations such as reproduction, crossover and mutation [2]. Applying these operators to previous population, genetic algorithm generates a new population. In the last generation, genetic algorithm selects the best solution of population as the best.

The purpose of this study is to design the minimum weight truss structures by using real value encoding in genetic algorithm. Real value encoding is discussed in several technical literatures, but no references have been found in the technical literature using real value encoding in genetic algorithm. The genetic algorithm program is coded in FORTRAN. This program includes displacement, stress and stability constraints. In the analysis of truss structures, matrix analysis of structures is used.

In real value encoding, the design of variables are represented by a single integer value, where the binary substring contains a set of integer value. For example, if a problem has three design variables, the example string contains three binary substrings, 100, 110 and 001, which sequence represents the integers 5, 7, 2, for each design variable, respectively. In real value encoding, the same substrings are represented directly by the integer values of 5, 7 and 2. In real value encoding, the length of strings is independent of the number of design variables and equal to the number of structural members or to group number. By means of this property, specially where the number of design variables and structure members are very much, value encoding make the genetic algorithm program fast and such kind of program is in need of very small computer memory when compared to the program including binary encoding.

The result obtained in this study showed that the program coded in this study is capable of efficient optimization of truss structures by using real value encoding in genetic algorithm.

1

Rajeev, S., Krishnamoorthy, C.S., Discrete Optimization of Structures Using Genetic Algorithms, Journal of Structural Engineering, 118 (5), 1992, 1233-1251. doi:10.1061/(ASCE)0733-9445(1992)118:5(1233)

2

Goldberg, D.E., Genetic Algorithms in Search, Optimization, and Machine Learning, 20th printing, Addison-Wesley Publishing Company Inc., New York, 1999.

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 £82 +P&P)