Session 10 Systems Engineering I

Friday, 13.11.2020, 11:00-12:00 o'clock

Presentations:


Michael Hillebrand: A design methodology for deep reinforcement learning in autonomous systems

11:00-11:20 o'clock

(Paper ID: 1190)

Autonomous systems such as mobile robots will play an important role in fields like industrial production, transportation or in hostile environments such as space. One of the most fundamental problem in autonomous mobile robotics is autonomous navigation. One new to solve this task is deep reinforcement learning. However, the application of deep reinforcement learning involves non-trivial design decisions. Previous work have failed to address a design methodology for deep reinforcement learning systems. In this paper, we propose design methodology and discuss design decisions for deep reinforcement learning in autonomous systems.

Paper PDF

Seyed Sina Shabestari: Decision support for Design Conflicts: A model-based method to analyze the interactions between technical requirements and product characteristics

11:20-11:40 o'clock

(Paper ID: 1104)

The diversity and multi-disciplinary of the technical products introduces conflicts as not all of the requirements can be fulfilled to the satisfaction of the stakeholders. Here the knowledge on the interactions between technical requirements and the product characteristics can help to find solutions for the optimum trade-offs between the requirements.

In this manuscript the state space of the product concept with the integrated technical requirements is explored. The generated data set is analyzed by the sensitivity analysis and later trade-offs are evaluated. As an example, the application of the proposed method in the development of a self-balancing robot is studied.

Paper PDF

Daniel Kloock-Schreiber: Application of System Dynamics for holistic Product-Service System Development

11:40-12:00 o'clock

(Paper ID: 1110)

In order to develop Product Service Systems (PSS), a holistic view on the system and a coequal development of service and product parts is necessary. Particularly for the beginning of the development of PSS, existing approaches show lacks and start with vague defined initial phases. This leads to inadequate methodological support for the consistent design of the overall system and simultaneous elaboration of the requirements down to the parameters of individual components. Therefore, a procedure is required that completely maps the PSS and enables detailed development for relevant individual areas, taking into account existing constraints. At the beginning of the development a model is necessary, which first defines the system boundaries of the PSS and maps the performance and control flows of the system. In addition, the integration of further actors into the PSS must be made possible.

This paper presents an approach that uses System Dynamics (SD) to design a PSS. With this approach, the representation of the system is initially possible at a high level of abstraction, whereby the representation can be further refined and detailed. Parallel to this, a preliminary design for planning and controlling media flows can be carried out from the first system representation and further detailed parallel to the system representation. An essential advantage is that the detailing can also only be carried out for individual areas, which can be displayed in sub-models, but can also be reintegrated into the overall representation.

The sub-models can be implemented function-specifically on the basis of resources and competencies of individual actors. For system-relevant areas, planning and design can be concretized in the sub-models (which can be realized by products as well as services) down to the lowest hierarchy level. This can take place up to the definition of individual physical component parameters and has thus up to the phase of the elaboration effects on the development of the parts. In return, the effects of changes in system-relevant parameters on the overall system can also be examined. For the PSS, a model is built in which system-determining functions and principles are represented and developed.

The model is constructed in such a way that non-system-determining functions and principles are defined as variables or black boxes. Requirements and parameters are derived from this system development. These are used for the further development steps in the development process. Depending on whether it concerns system-relevant areas or not, the entry into the development process takes place later in the elaboration phase (e.g. in the area of detailed design) or partly earlier in the concept phase (e.g. function development). It is also possible to enter an early phase in the development process of the individual parts, accompanied by already defined functions, sub-functions or parameters that must not be changed in the course of development. With this approach a holistic development of the system with all product and service parts as well as their connections and dependencies is possible.

Paper PDF

Steve Zimmer: Systematic engineering of functionally integrated wireless power transfer systems for electric vehicles

12:00-12:20 o'clock

(Paper ID: 1164)

This paper presents a systematic approach to the functional integration of an on-board unit of wireless power transfer systems of electric vehicles into an underbody cover. On-board units of wireless power transfer systems consist of macro-scaled elements with high functional interdependencies. The integration of these elements requires the multidisciplinary end-to-end consideration of the cross-domain and functional interdependencies. To manage this complexity, a problem-specific and purpose-oriented approach is required. Based on the existing V-Model of VDI 2206, the proposed approach considers simultaneously the domains of electronics, mechanics, production and assembly within the development process.

In line with the approach, this work presents the conception and evaluation of three functionally integrated on-board units. These concepts provide a reduced vertical dimension. This improves significantly the interoperability of wireless power transfer systems.

Paper PDF

Karl Kübler: Towards an Automated Product-Production System Design − Combining Simulation-based Engineering and Graph-based Design Languages

12:20-12:40 o'clock

(Paper ID: 1180)

In this paper, the authors elaborate a combination of graph-based design and simulation-based engineering into a new concept called Executable Integrative Product-Production Model (EIPPM). Today, the first collaborative process in engineering for all mechatronic disciplines is the virtual commissioning phase. Therefore, Digital Twins (DT) are modeled and run in a simulation. The authors see a hitherto untapped potential for the earlier, integrated and iterative use of DTs in a simulation-based engineering for the development of production systems. Seamless generation of and exchange between Model-, Software- and Hardware-in-the-Loop simulations is necessary. Feedback from simulation results will go into the design decisions after each iteration.

The presented approach combines knowledge of the domain “production systems technology” together with the knowledge of the corresponding “product” using a so called Graph-based Design Language (GBDL). Its central data model, which represents the entire life cycle, results of an automatic translation step in a compiler. Since the execution of the GBDL can be repeated as often as desired with modified boundary conditions (e.g. through feedback), a design of experiment is made possible, whereby also unconventional solutions are considered.

The novel concept aims at the following advantages: Consistent linking of all mechatronic domains through a data model (graph) from the project start, automatic design cycles exploring multiple variants for optimized product-production system combinations, automatic generation of simulation models starting with the planning phase, feedback from simulation-based optimization back into the data model.

Paper PDF

Discussion and Comments