| This research introduces a Petri Net based approach for conflict detection in multi-agent systems (MAS). Conflict detection approaches are either static or dynamic. Static approaches detect the conflicts during multi-agents' integration, and dynamic approaches detect the conflicts during run-time. The Petri Net based approach is an extension of the goal structure tree and the E-PERT diagram. This approach intends to keep the functions from both the goal structure tree and the E-PERT diagram and extends them with the advantages of Petri Net. It also keeps mathematical analysis characteristics of existing visualized approaches in the static analysis perspective and extends them with the abilities of simulation and real-time controller implementation in the dynamic analysis perspective. The interactions among agents were also defined, and three types of conflicts, goal, plan, and resource, can be detected.; Conflict resolution is a challenging issue due to the fact that it is generally difficult to analyze conflict quantitatively with multiple perspectives being considered. In this research, the researchers also extend the Petri Net based approach and propose a data envelopment analysis (DEA) based approach to analyzing conflicts, called conflict severity analysis. Conflict severity index (SI) is derived for each agent and can be further applied to develop an agent negotiation mechanism. Overall, the DEA based approach offers two potential advantages. First, both qualitative and quantitative data can be analyzed simultaneously. Second, it applies a mathematical mechanism to the problem, with the potential of being relatively straightforward to computerize.; This research further proposes a distributed approach for conflict resolution. The potential advantages of the distributed approach are to propose the strategies for replacing the interactions during a conflict, to present an interaction-based centralized algorithm to resolve the conflicts statically, and to present a negotiation mechanism to resolve the conflicts dynamically. |
