INGAR   05399
INSTITUTO DE DESARROLLO Y DISEÑO
Unidad Ejecutora - UE
congresos y reuniones científicas
Título:
An agent-based simulation model using decoupled learning rules to (re)schedule multiple projects
Autor/es:
ERNESTO MARTINEZ; LAURA TOSSELLI; VERÓNICA BOGADO
Lugar:
La Plata
Reunión:
Congreso; CACIC 2017-Congreso Argentino de Ciencias de la Computación; 2017
Institución organizadora:
Universidad Nacional de La Plata. Facultad de Informática
Resumen:
Competitive pressures and business globalization have led manyorganizations, mainly technology-based and innovation-oriented companies, toadopt project-based organizational structures. In a multi-project context withinenterprise networks, reaching feasible solutions to the multi-project(re)scheduling problem represents a major challenge, where autonomy anddecentralization of the environment favor agent-based simulation This workpresents and validates a simulation-based multi-agent model using the fractalcompany concept to solve the complex multi-project (re)scheduling problem inenterprise networks. The proposed agent-based model is tested trough a set ofproject instances that vary in project structure, project parameters, number ofresources shared, unplanned events that affect them, etc. Results obtained areassessed through different scheduling goals, such project total duration, projecttotal cost, leveling resource usage, among others to show that decoupled learningrules allows finding a solution which can be understood as a Nash equilibriumfor the interacting agents and it is far better compared to the ones obtained withexisting approaches.