A novel hybrid approach for supply chain modeling and analysis.

MELE, FERNANDO D.; MÉNDEZ, CARLOS A.; ESPUÑA, ANTONIO; PUIGJANER, LUIS

Barcelona, España

Congreso; 10º Congreso Mediterráneo de Ingeniería Química; 2005

Sociedad Española de Química Industrial e Ingeniería Química

In this work, a multi agent-based simulator is used to represent the system dynamics responding to uncertainties on-line through local dispatching rules, invoking when necessary local optimization modules, and solving conflicts through message exchange. A problem ever arising when representing SC networks using software agents is the nature of the tools employed to model the different decisions the central agent has to make. Usually, these decisions are made on the basis of if-then rules or techniques belonging to the field of the so-called artificial intelligence. Thus, it is proposed to use a hybrid system which combines the modeling capabilities of the multi-agent systems together with the optimization strengths of relatively small mathematical programming models to solve in a more efficient and integrated manner the decision problems that the central agent faces along the simulation run.