Genetic algorithm metaheuristic in the solution of forest management models

Abstract

This work aimed to test the Genetic Algorithm (GA) metaheuristic evaluating its effectiveness and efficiency in the solution of this kind of problem, and comparing its results with those obtained by the software CPLEX. A completely randomized design, as the factorial arrangement, was used to analyze the effect of different parameters on GA performance. Sizes of initial population (Pini), crossing-over rates (Tcross) and crossing-over methods breaks  (Mcross) were the analyzed factors. In the cases that the interactions were significant by F test (P < 0.05) the differences among the means were tested by Tukey test at 5% probability level. The percentage distance (distance between the answer of AG and the exact answer) and the time of processing were used as a measure of effectiveness and efficiency, respectively. The initial population is the factor that most influenced the AG performance considering percentage distance and processing time, so that for larger Pini are found greater proximity of GA response to the exact response and also greater processing times.