A heurística da razão-R aplicada a problemas de gestão florestal

Abstract

The R-ratio heuristic method for solving large scale forest planning problems was analyzed according to Artificial Intelligence principies The method is promising ifconsidered ali relevant aspects related to the construction of efficient searches, mainly due to its simplicity and consistent logic The analysis based on Artificial Intelligence introduced some changes to the original implementation. These improvements were tested in problems previously solved by older versions of the heuristic, allowing further comparisons. The improvements were evaluated based on effíciency, efficacy and computational time. Results show that effíciency improved for ali problem dimensions and time was reduced about 40 times. The effíciency, measured by concepts derived rrom heunstic search testing criteria, reveals that the R-ratio heuristic is very efficient even for large problems. Local optimum escape strategies were tested and the respective efficient intervals for the heuristic parameters were also defined. Interruption criteria ofthe search for less productive strategies were ínvestigated. And finally, the results were discussed under the R-ratio buildins logic and the Artificial Intelligence assumptions, that justify the eood performance ofthe heuristic.