The subprojects of project area B aim at developing design methods and techniques for self-optimizing systems. Self-Optimizing systems are built as multi-agent systems. Consequently, specification- and modeling techniques for the agents as well as the multi-agent system are required. Support for formal and experimental quality analyses is an important characteristic of these techniques. The design methodology assists the integration of the design activities of the participating domains.

Software Design Techniques

Subproject B1 develops a multi agent modeling paradigm for self-optimizing systems. The proposed modeling techniques are used for the specification autonomous agents and their communication. Building high-quality dependable software systems is the major driver for the development of the modeling languages. The integration of mechatronic elements requires that their specific properties are properly addressed in the structural specification, the definition of interfaces, the specification of coordination activities and definition of quality criteria. In the first project phase, an approach has been developed which is based on domain specific design patterns and compositional verification activities. In the second phase, we extend our approach for specification of hybrid agents with discrete and continuous behavior. Online reconfiguration of self-optimizing systems is another area of interest under consideration of the safety and fault-tolerance aspects of the mechatronic domain.

Design methodology

Subproject B2 integrates the gained knowledge of all subprojects in the CRC to a design instrument for s.o. systems, enabling third persons to successfully develop s.o. systems. The instrument covers five substantial fields: 1) Design processes, 2) design methods, 3) specification techniques, 4) a design environment, which enables computer aided design, as well as 5) a knowledge base. Emphasis of the first promotion period was a procedural model for the early phases "Planning and clarifying the task" and  "Conceptual design" of the design process as well as the domain-spanning specification of the principle solution. It appeared that the effect paradigm of self-optimization leads to a
comprehensive advancement of the designing methods of modern mechanical engineering. In the second promotion period the work within the five mentioned fields will be continued. The early phases with the result "Principle solution" will still be the centre of attention, because here the crucial decisions take place. A substaintial goal is to enhance and formalize the coherent system of partial models of the principle solution. On this basis the specification activties of the remaining subprojects, which take place after the conceptual phase, shall tie up, first computer aided functional verification shall take place and the computer aided process management in the early phases shall be substaintially improved. Furthermore self-optimization requires new methods to identify the potential of S.O. and the S.O. functionality which can be realized.

Virtual Prototyping

Subproject B3: It is hardly possible for developers to fully anticipate the behavior of a self-optimizing mechatronic system at the design stage. This is why we need processes, methods and tools that can perform rational explorations to render the mechanisms of the self-optimization visible and intelligible. This subproject’s higher-level objective is therefore to develop and test new interaction, representation and analysis techniques for the design and validation of self-optimizing systems using virtual prototypes. The core of our work during the initial funding phase was to develop an innovative synthesis/analysis environment based on the interactive assembly of solution elements and subsequent generation of a computer-based representation of the construction, the active structure and the control system. We also developed important modules of a simulation and visualization infrastructure. The higher-level objective of the subproject B3 during the period covered by the second grant application is to set up virtual test environments to support the engineers through all the processes of refining the requirements and systematic testing, right down to a readily intelligible presentation and explanation of the self-optimization processes using virtual reality and augmented reality.

Subproject B1 develops the modeling techniques for mechatronic multi-agent-systems. Subproject B3 assists the engineer to comprehend complex behavior and enables a rational exploration of the modeled system. Both subprojects support the verification and validation activities which are highly important during development of self-optimizing systems. Subproject B1 develops methods for the automatic validation of those systems. In contrast, subproject B3 develops methods for experimental validation. The cooperation of subprojects of the project area B is presented in the left picture. According to it subproject B2 forms the framework for the comprehensive design instrument and prepares necessary contents after the shown structure (five fields). Orthogonal to these five filds are two particularly important aspects, which result from the nature of the complex data processing of s.o. systems and are each treated in one subproject.