Behaviour adaptation in the multi-agent, multi-objective and multi-role supply chain

摘要

Researchers, practitioners and enterprise software providers are realising the potential of agent-based technology to automate supply chain procurement to achieve consistent, traceable decision making. Asthe complexity of supply chains grow, these systems will gain more attention. In this paper, we model and simulate a complex autonomous supply chain managed by computational agents that aim tominimise lead time and maximise revenue through evolutionary multi-objective optimisation. The agents are in a competitive environment where they take on the roles of both client and producer. Inaddition to optimising their production strategy, they also have the opportunity to dynamically finetune their decision parameters when it comes to selecting their own suppliers, using the AnalyticalHierarchy Process. It is observed that computational agents are capable of functioning in such complex environments, effectively converging to policies in synergy with their market. Multi-objective, multiroleoptimisation results in er overall supply chain performance than tests where agents have singleobjectives and single-roles. Our study forms an exploratory step towards more realistic agent-based supply chains where analytical methods are unavailable.