| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 4.85 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
Atualmente a gestão de projetos implica a tomada de decisões face a problemas cada vez mais complexos, com mais variáveis e com mais objetivos a otimizar e/ou a cumprir que, por sua vez, são sistematicamente conflituosos entre si. Neste âmbito, surge a proposta de um framework composto por um método de otimização multiobjetivo e métodos multicritério de apoio à decisão que, numa primeira instância, permitem atribuir pesos aos critérios e, seguidamente, permitem ordenar as soluções, de forma a suportar a tomada de decisão por parte do gestor do projeto. O Arrefecimento Simulado Multiobjetivo foi o método de otimização multiobjetivo utilizado, pela sua simplicidade e versatilidade, tendo-se estudado o impacto da incorporação de duas estratégias de pesquisa (regresso ao arquivo e grelha exploratória), sendo que ambas contribuíram para a melhoria do desempenho do algoritmo e, consequentemente, para determinar as soluções de compromisso. A Entropia de Shannon e o AHP foram os métodos utilizados para a atribuição de pesos aos critérios, sendo que a primeira metodologia deverá ser aplicada por decisores mais inexperientes (uma vez que é objetiva e se baseia na informação existente) e a segunda permite uma ponderação mais subjetiva e flexível dos pesos dos critérios, com base na experiência do gestor de projeto. A atribuição dos pesos permite a utilização de métodos de ordenamento de soluções, como o TOPSIS e o VIKOR, que tendo o mesmo objetivo originam rankings diferentes. Com o objetivo de tornar o ranking das soluções mais robusto, sugere-se um método híbrido que conjuga os scores das duas metodologias, destacando as soluções que são melhores em ambos os métodos. A metodologia proposta permite resolver problemas de otimização multiobjetivo em problemas de trade-off, suportando as decisões do gestor de projeto. Contudo, as ferramentas utilizadas encontram aplicações em diversas áreas.
Nowadays, project management implies the decision-making regarding problems with increased complexity, with more variables and objectives to optimize and/or to fulfil, which are conflicting with each other. In this context, it is proposed a framework composed by a multiobjective optimization method and multicriteria decision-making methods, which at first, allow for the weight attribution and, secondly, to rank the solutions allowing the project manager to support its decision. Multiobjective Simulated Annealing is the multiobjective optimization method chosen, due to its simplicity and versatility, allowing the study of the incorporation of two search strategies (return to archive and exploratory grid), which improved algorithm performance, increasing the number of non-dominated solutions found. Shannon’s entropy and AHP were the chosen methods for weighting criteria. Junior project managers should choose the first method (since it is objective and based in the existing information) and the latter is more subjective and flexible, relying on the decision’s experience. Weighing allows the ranking of the compromise solutions, like TOPSIS and VIKOR, which, although they the same objective, produce different rankings. In this context, an hybrid method combining both method were proposed to deliver a more robust ranking method, highlighting the best solutions from both methods. The methodology proposed allows the resolution of multiobjective optimization regarding trade-off problems, supporting the project manager’s decisions. However, the methods used might be used in many different fields.
Nowadays, project management implies the decision-making regarding problems with increased complexity, with more variables and objectives to optimize and/or to fulfil, which are conflicting with each other. In this context, it is proposed a framework composed by a multiobjective optimization method and multicriteria decision-making methods, which at first, allow for the weight attribution and, secondly, to rank the solutions allowing the project manager to support its decision. Multiobjective Simulated Annealing is the multiobjective optimization method chosen, due to its simplicity and versatility, allowing the study of the incorporation of two search strategies (return to archive and exploratory grid), which improved algorithm performance, increasing the number of non-dominated solutions found. Shannon’s entropy and AHP were the chosen methods for weighting criteria. Junior project managers should choose the first method (since it is objective and based in the existing information) and the latter is more subjective and flexible, relying on the decision’s experience. Weighing allows the ranking of the compromise solutions, like TOPSIS and VIKOR, which, although they the same objective, produce different rankings. In this context, an hybrid method combining both method were proposed to deliver a more robust ranking method, highlighting the best solutions from both methods. The methodology proposed allows the resolution of multiobjective optimization regarding trade-off problems, supporting the project manager’s decisions. However, the methods used might be used in many different fields.
Description
Keywords
Gestão de Projetos Otimização Multiobjetivo Métodos de Decisão Multicritério Entropia de Shannon Arrefecimento Simulado Multiobjetivo AHP TOPSIS VIKOR Project Management Multiobjective Optimization Multicriteria Decision Methods Shannon’s entropy Multiobjective Simulated Annealing