The objectives of this work was to develop and test a Genetic Algorithm (GA) to solve problems of forest management with integer constraints. GA was tested in five problems containing 93 - 423 decision variables, periodically subject to singularity constraints, minimum and maximum production. The problems had the objective of maximizing the net present value. GA was codified into delphi 5.0 language and the tests were performed in a microcomputer AMD K6II 500 MHZ, with RAM memory of 64 MB and hard disk of 15GB. The GA performance was evaluated according to the efficacy and efficiency measures. The different values or categories for the GA parameters were tested and compared in relation to their effects on the algorithm efficacy. The selection of the parameters' best configuration was performed by using the L&O test at 1% probability and analyses via descriptive statistics. The parameters' best configuration provided for GA average efficacy was of 94.28%, minimum value equal to 90.01%, maximum value equal to 98.48%, with coefficient of variation of 2.08% of the mathematical optimum, obtained by the exact algorithm branch and bound. As for the larger problem, the efficiency of GA was five times superior to the efficiency of the exact algorithm branch and bound. GA was found to be a quite attractive approach to solve important forest management problems.
Forest management; metaheuristics; genetic algorithm
