Generic Need Estimating Agents for Resources Forecasting
July 13, 2012
Posted by on
Ayda KADDOUSSI, Slim HAMMADI, Emmanuel DUFLOS, Philippe VANHEEGHE and Hayfa ZGAYA
The management and optimization of the crisis management supply chain is very complex and involves multiple concepts: supply resources, probability and statistics, means of transportation, disturbances on the delivery dates, and the importance of decisions involving human lives. To meet these requirements, multi-agent systems are a well-suited solution for modeling the supply chain through interactive autonomous entities. Our model represents a distributed logistic system in which flows of resources are hierarchically forwarded from one zone to another, taking into account the randomness of resources’ consumption. The issue is then to optimize the procurement policy to avoid shortages that could cripple the whole system. In this paper we propose the multi-agent technology for modeling the different actors of the logistic chain. Then we propose an innovative method to estimate the future needs in resources for every zone. Our forecasting method is based on fuzzy calculations combined to ARMAX time series modelling. The objective of our work is to avoid, in a crisis situation, stock outs by optimizing the estimation of the future needs in resources for each zone and balancing the flows throughout the system.
Full Paper in PDF Document