Towards a multi-agent dynamic supply chain simulator for analysis and decision support

Having effective decision support tools for coordinated operation of supply chains (SC) operating in a vague and uncertain environment is essential for success in the highly competitive global market. Supply chains involve the activity and interaction of many entities, which have to make coordinated strategic and managerial decisions in a highly dynamic fashion. Even in the data rich environment emerging with the advances in the information technology field, uncertainty still prevails due to conflicting information and inconsistent data. Integration of soft computing and agent technologies provides decision support for optimal decision making in SC design and management in uncertain environment. In this paper, we describe a multi-agent supply chain simulator (MASCS) developed to study novel approaches to dynamic SC configuration and control using different soft-computing technologies: reinforcement learning, fuzzy rules and perceptual forecasting. The simulator permits the analysis of the different segments of a SC, also can be used at the enterprise level to model just-in-time production processes. We describe a case study developed to obtain semi-optimal SC configurations based on different demand patterns.