3.3. Multi-Agent System for the autonomous negotiation between maintenance requirements and available resources
The ability of a considered system to assess its own state in order to ask for maintenance activities is the key
element to enable autonomous planning and control within the general socioeconomic maintenance system.
Nevertheless, the processing of maintenance tasks is limited through specific resources, accessibility, available spare
parts, tasks with higher priority, etc. These actors of the maintenance system also follow internal rules and decision
logics, i.e. the mitigation of costs or risks. In conclusion, entities of the socioeconomic maintenance systems require
the ability to negotiate possible constellations according to their local objectives. In complex and dynamic
environments this promises to gain efficiency also on the macro level because in cooperating systems the entities
pursue a globally optimized behavior and want to achieve common goals.
Focusing on a multi-agent-approach in order to enable this kind of negotiation and cooperating system, the
different entities of the overall system have to be modeled as intelligent agents, which are able to act without the
intervention of humans or other systems [20]. Each agent is self-content and acts autonomously in order to enable
autonomous decision-making. Within the proposed multi-agent system the planning and control is shifted from a
central system with hierarchical structures to decentralized autonomously acting agents. The general problem is split
into smaller problems that agents solve locally.
Regarding the described case of maintenance management, the following Table 1 gives an overview of possible
agent layers that model the corresponding behavior of entities.
338 Marco Lewandowski and Stephan Oelker /