Aplicação da técnica de recozimento simulado em problemas de planejamento florestal multiobjetivo

Abstract

This study has as objective to discuss the Simulated Annealing's applicability in the solution of problems of planning forest under multiple objectives. Therefore, an optimization model is proposed under multiple objectives and implemented an algorithm for the solution of the model, PSA (Pareto Simulated Annealing), a variation of Simulated Annealing simple objective, for the solution heuristic of the model under the vision of multiple objectives. Later on, various tests were deed with the algorithm in problems of several loads to measure its efficiency and its efficacyand compared the results with the CPLEX, a software that uses an exact algorithm for the solution of problems of Programming Integer 0-1, and with the algorithm SPEA implemented by Teixeira (2002), that uses the metaheurístic Genetics Algorithms for the solution of the same problem. In terms of efficacy between SPEA and PSA can be affirmed that the two were equivalent for problems of small load; the efficiency of PSA, in relation to SPEA, was superior. For problems of medium to great load SPEA was not able to solve probably for loss of memory of the system, while PSA showed a good performance. In a competitive environment where the taking of decision should be made with agility, PSA showed interesting in spite of not being an exact algorithm, it obtained an approximation of 99,68%, in a worthless time when compared withCPLEX. It is concluded by the applicability and for the efficiency of the metaheurístic Simulated Annealing in the solution of problems of planning forest under multiple objectives and, therefore, for the applicability of PSA, in particular, as a good alternative for the treatment of these problems. Its performance showed satisfactory and very good in relation of the algorithms evolutionary, in particular SPEA, on theproblem-test group, where several parameters were defined through of real data associated with a company of the forest area.