Nonlinear Mixed Variable Optimum Design Applying Adaptive Range Genetic Algorithms.
Masao ARAKAWA, Ichiro Hagiwara · TRANSACTIONS OF THE JAPAN SOCIETY OF MECHANICAL ENGINEERS Series C · 1998
In design problems, it is possible to classify design variables into three classes: integer, discrete and continuous variables. It is said that mixed variable optimization is the one of the most daunting problems in design optimization. Genetic algorithms are very effective algorithms especially for large scaled combinatorial optimization problem. Thus, they are also effective in integer and discrete programming. However, as they are not well coded to its tedious expression in converting chromosomes to design variables, we need to do some special efforts to overcome these flaws. In the proposed method, it automatically adapts a searching range according to the situation of the generation. Thus, we do not have to take tedious expression into account. Moreover, we do not have to give too many genes to chromosome, thus not only we can save computational memory but also the convergence becomes more better. In this paper, we combine the proposed integer and discrete adaptive range genetic algorithms and adaptive real range genetic algorithms which we presented in the former studies, and present an extended genetic algorithms method which adapts a searching range for each design variable for mixed integer, discrete and continuous programming. We applied the proposed method to well known test problems in mixed optimization problems, compare the results with the other methods and show the effectiveness of the method.