In order to compete in challenging global markets, Rolls-Royce must have the capability to manage complex processes effectively. As the environment surrounding these processes is constantly changing, the ability to adapt to meet new requirements is essential. The challenge to the business is to quickly adopt an appropriate course of action for each set of circumstances, anticipated or otherwise. An individual decision-maker faced with this situation will normally have several options available. The difficulty is assessing the ramifications of each within a given time frame. To improve the quality of this evaluation, computerized decision support tools can be used. Such tools offer the ability to assess a multitude of options in a short space of time, using a combination of expert knowledge and real data. Having developed an extensive aero-engine service business, Rolls-Royce is required to maintain a highly dependable aftermarket infrastructure. Therefore, the ability to compare how any one particular aftermarket design would perform relative to another is an essential capability. Agent-based systems offer an approach that is both intuitive and interactive, modelling individual entities in the system from the bottom up, capturing low-level interactions that ultimately determine the overall performance of the system. This provides the flexibility and transparency to allow trustworthy analysis and evaluation to take place. To address this business need, an agent-based aftermarket model has been developed. An agent-based system is made up of small software programs built to operate just like a human team. Each agent has a set of capabilities and knowledge, but must work with other agents to achieve the overall goal. Agents can react to changes, adapting and re-planning if a better approach is identified. This paper describes how the model was constructed and the resulting analysis that it facilitates. The model has the ability to replicate the likely service characteristics that would be in place during the full lifecycle of the product. All aspects of the necessary infrastructure are captured, based upon the roles and capabilities of the constituent elements. Agents are used to represent a variety of objects and functions, including airports, airlines, aircraft, overhaul facilities and logistics. Through planning and negotiation, representative decisions are made by these agents to determine when an overhaul should take place and what the workscope should be based upon defined policies. A large selection of configurable parameters can be set by the user to accurately reflect the proposed scenario, providing a powerful what-if analysis tool that can be used to drive the design process, ensuring that product attributes and performance are aligned with the available maintenance infrastructure.
Multi-criteria scheduling: an agent-based approach for expert knowledge integration
