A multi-agent methodology for multi-level modeling of mechatronic systems

• We propose a multi-agent methodology for multi-abstraction modeling of mechatronic systems.
• We decompose the design into agents linked with relationships and managed with design rules.
• The methodology is applied to a design case of energy harvesting system.
• We illustrate how this approach ensures continuity between abstraction levels.
• We also illustrate how designers identify modeling errors using checking rules.

Mechatronic design aims to integrate the models developed during the mechatronic design process, in order to be able to optimize the overall mechatronic system performance. A lot of work has been done in the last few years by researchers and software developers to achieve this objective. However, the level of integration does not yet meet the purposes of mechatronic system designers, particularly when dealing with modeling changes. Therefore, new methodologies are required to manage the multi-view complexity of mechatronic design. In this paper, we propose a multi-agent methodology for the multi-abstraction modeling issue of mechatronic systems. The major contribution deals with proposing a new method for the decomposition of the multi-level design into agents linked with relationships. Each agent is representing an abstraction level and both agent and relationships are managed with rules. By considering an application to a piezoelectric energy harvesting system, we show how we associate agents, rules and inter-level relationships to multi-abstraction modeling. We also show how modeling errors are identified using this approach.