Simulação multiagentes aplicada ao planejamento da produção florestal sustentável

• Ano de defesa: 2012
• Autor(a) principal: Araújo Júnior, Carlos Alberto
• Orientador(a): Leite, Hélio Garcia
• Banca de defesa: Ribeiro, Carlos Antônio Alvares Soares , Nogueira, Gilciano Saraiva
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Viçosa
• Programa de Pós-Graduação: Mestrado em Ciência Florestal
• Departamento: Manejo Florestal; Meio Ambiente e Conservação da Natureza; Silvicultura; Tecnologia e Utilização de
• País: BR
• Palavras-chave em Português: Regulação florestal; Inteligência Artificial Distribuída; Heurística
• Palavras-chave em Inglês: Forest regulation; Distributed Artificial Intelligence; Heuristics
• Área do conhecimento CNPq: CNPQ::CIENCIAS AGRARIAS::RECURSOS FLORESTAIS E ENGENHARIA FLORESTAL::MANEJO FLORESTAL
• Link de acesso: http://locus.ufv.br/handle/123456789/3105

This study aimed to evaluate the application of the paradigm of Distributed Artificial Intelligence in forest management problems. For this, we developed two Multi-Agent Systems: one for the long-term planning of a forest schedule (i) and other for the harvester planning in one year (ii). For the MAS (i) were modeled three types of agents: harvest agent, inventory agent and control agent. The environment was a forest company with 120 stands in area equal to 4.269,29 ha. The agents acted in synchrony in the environment trying to achieve the overall goal of the system, that was to determine the sequence of annual harvest to return higher net present value. We considered constraints about integrity of stands, minimum and maximum wood s demands, minimum and maximum areas managed and harvested annually with priority given to older forests. The MAS was able to generate feasible scenarios and evaluate which of these was the best solution. The MAS (ii) considered the modeling of three kinds of agents: registration agent, harvest agent and control agent. The focus of MAS (ii) was to generate plans that indicate the stands to be harvested in each month in the planning horizon of one year. Were consider the restrictions of variation in monthly production that were less than 10% and variation in the average density of the wood delivered to the processing unit less than 5%. For this MAS was assessed the behavior of the same in both situations. In the first case the agent had a collection drive for sequencing of farms to be managed and in the second situation this knowledge was stripped. The MAS was able to generate cutting scenarios that met the restrictions. Increasing the degree of knowledge of the agent in relation to farms sequence for moving the harvest module increased the system performance. It was concluded that Multiagent Systems can be used as a tool to plan the production of forest in long and medium term.