A Design Methodology for Self-Optimizing Systems

Innovative self-optimizing systems which go far beyond current approaches for mechatronic products become possible when systems are enabled to optimize their own behavior at runtime. Such self-optimizing systems are characterized by their ability to endogenously modify their objectives in response to changing conditions and autonomously adapt their parameters and structure and as a result their behavior to fulfill their objectives. This paper outlines a systematic approach for the development of self-optimizing systems. The approach helps to reduce the considerable additional development efforts resulting from self-optimization by employing different forms of patterns throughout the whole development process to enable the reuse of design knowledge. At first, the employed notion of patterns covers the multiple disciplines involved as well as different phases of the development process. In addition, the patterns are used to enable a systematic transition between the different milestones of conceptual design such as the function hierarchy, the active structure, and the construction and component structure. The approach is presented using the example of autonomously driving shuttles which self-optimize their behavior.