This work aimed to develop and test an algorithm based on Tabu Search (TS) metaheuristics; to solve problems of forest management with integer constraints. TS was tested in five problems containing between 12 and 423 decision variables subjected to singularity constraints, minimum and maximum periodic productions. All the problems aimed at maximizing the net present value. TS was codified into delphi 5.0 language and the tests were performed in a microcomputer AMD K6II 500 MHZ, RAM memory 64 MB and hard disk 15GB. TS performance was evaluated according to efficacy and efficiency measures. The different values or categories for the TS parameters were tested and compared to their effects on algorithm efficacy. The best configuration of the parameters was selected by applying the L&O test at 1% probability and analyses via descriptive statistics. The best parameter configuration provided TS an average efficacy of 95.97%, a minimum value equal to 90.39%, a maximum value equal to 98.84%, with a variation coefficient of 2.48% of the mathematical optimum. For the most complex problem, TS efficiency of the exact algorithm branch and bound. Thus, TS is a rrather attractive approach to solving important forest management problems.
展开▼
