No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
The autogenetic design theory and its practical application
Abstract
The main focus of the paper is to present the autogenetic design theory (ADT) as an evolutionary view of the design process. The ADT is an approach that transfers procedures from the natural evolution into the field of product development. The main thesis of the ADT is that natural evolution and the process of developing products are mainly similar. In order to fulfil requirements and boundary conditions of any kind (that may change at any time), both processes look for appropriate solution possibilities in a certain area, and try to optimise those that are actually promising (in an iterative way) by varying parameters and combinations of these solutions. In the natural evolution, these parameters are the organic bases, in the field of product development, these are the design parameters of the product. This paper gives an introduction to the ADT and presents one aspect (the description of solution space) in more detail. It also presents two practical examples. The first example shows, how a subset of methods of the ADT were successfully applied to improve a product. The second example shows, how even complex problems can be prepared for an optimisation using methods from the ADT.
... To reduce the risk of missing a good solution we developed the Morphix framework, a system to support conceptual design by combining the morphological box and the Autogenetic Design Theory [Kittel et al. 2011a] by using NOA, a framework that contains several Genetic Algorithm objects to support the product development process. Genetic Algorithms are very suitable for supporting conceptional design, because they can deal with different kinds of parameters and the objective function doesn't have to be continuous. ...
... The Autogenetic Design Theory (ADT) is an approach that transfers procedures from natural evolution to accomplish a broad description of product development and its processes, requirements, boundary conditions, and objects (including their properties) ( [Bercsey and Vajna 1994], [Vajna et al. 2005], [Kittel et al. 2011a]). The ADT is not another variety of Bionics. ...
... NOA is a framework that contains several GA objects, which can be chosen and configured by the user. It was applied and verified in various projects with different partners to support the embodiment design of products at different levels of detail ( [Mack et al. 1999], [Vajna et al. 2006], [Kittel et al. 2011a]). In order to reflect the various requirements and conditions to be fulfilled by the emerging product, the ADT solution space contains taboo zones that can't be entered by possible solutions. ...
In this paper we introduce Morphix, a framework to support the conceptual design by combining the morphological box and the Autogenetic Design Theory using a Genetic Algorithm, to reduce the risk of missing a good solution while using a morphological box.
Morphix generates concepts and the designer rates the proposed concepts by predefined or custom evaluation criteria. The framework was applied in different use cases. We found that the application of Morphix supports thinking "out of the box" and getting new insights on the product since the genetic algorithm regards the whole solution space
... This framework helps predict that the project's development process scope will probably change, due to the status of design, new solutions, dynamically changing customer, or market requirements. That is why the definition and description of target functions happen at the start of the project [7]. [3,4] During product development, project team members are focusing on operative tasks to deliver necessary results. ...
This paper presents similarities and connection points of Autogenetic Design Theory (ADT) and evolutionary methods. During a short historical overview, ADT is positioned in a timeline. The authors show where common issue points between the two approaches are. Then demonstrate how to use these methods, to avoid a long progress of iteration and tests of several versions and directions. Via an example study of truck floor design, benefits of evolutionary methods are highlighted and made understandable for readers.
Die vorliegende Dissertation beantwortet die Forschungsfrage der effizienten Bearbeitung von Optimierungsaufgaben in der Produktentwicklung durch den Einsatz von Parallelisierungsmethoden unter Berücksichtigung der verfügbaren Ressourcen innerhalb einer Organisation. Basierend auf Analogiebetrachtungen zum Verhalten interdisziplinärer Entwicklungsteams wird eine Methode zur effizienten Parallelisierung von Evaluationsprozessen von Optimierungen entwickelt und prototypisch in einem Framework umgesetzt. Unter
den Begriffen Concurrent Optimization und Simultaneous Optimization werden die zur Parallelisierung von Produktentwicklungsprozessen eingesetzten Methoden des Concurrent Engineering und des Simultaneous Engineering auf die Anwendung in Optimierungsverfahren übertragen.
Die Parallelisierung der Evaluationsprozesse basiert auf der Dekomposition des Optimierungs-und des Evaluationsmodells und erfolgt innerhalb des entwickelten Frameworks durch dynamische Allokation. Da die Bearbeitung der Prozesselemente eines Evaluationsprozesses dabei nur auf der Verfügbarkeit und der Stabilität der benötigten Informationen und Ressourcen basiert, wird eine hohe Flexibilität in der Anwendung sowie die dynamische Reaktion auf spontane Veränderungen der Ressourcen ermöglicht. Als Ressourcen dienen die Einheiten eines Workstation-Clusters. Dies ermöglicht die Nutzung freier Rechenkapazitäten vorhandener Workstations und stellt somit eine kosteneffiziente verteilte IT-Umgebung zur Verfügung. Weiterhin werden die vorhandene Software und die verfügbaren Lizenzen berücksichtigt. Durch die prioritätenbasierte Bearbeitung der Prozesselemente kann die Effizienz einer Optimierung zusätzlich erhöht werden, da somit unnötiger Leerlauf vorhandener Ressourcen vermieden wird. Die durchgeführten Untersuchungen zeigen, dass Optimierungsverfahren im Vergleich zu Heuristiken die wesentlich effektivere und flexiblere Lösung zur Ermittlung der optimalen Prioritätswerte darstellen. Der genetische Algorithmus NSGA-II generiert dabei durchgängig die optimalen Prioritätswerte, welche in der kürzesten Laufzeit der Produktoptimierung resultieren.
Die Validierung des Frameworks erfolgt anhand von drei Fallstudien, welche charakteristische multidisziplinäre Optimierungsaufgaben aus realen Entwicklungsprojekten darstellen. Dies spiegelt ein industrienahes Umfeld wider, in dem freie Workstations während Pausen, in der Nacht oder am Wochenende als Ressourcen in einem Cluster zur Verfügung stehen können. Das entwickelte Framework kann sowohl in homogenen als auch in heterogenen Workstation-Clustern eingesetzt werden, welche kostenneutral aus einer beliebigen Anzahl in einem Netzwerk verbundener Workstations realisiert werden können. Zusätzliche Hardware wird nicht benötigt. Zudem besteht keine Notwendigkeit eines gemeinsamen Speichers, da der Datentransfer direkt zwischen den Workstations erfolgt. Jede zur Verfügung stehende Workstation kann dem Cluster schnell und einfach hinzugefügt oder aus dem Cluster entfernt werden, auch während laufenden Optimierungen. Die Workstations des Clusters stehen daher jederzeit zur interaktiven Arbeit zur Verfügung und die interaktive Arbeit wird nicht beeinträchtigt. Das entwickelte Framework eignet sich daher für die Verwendung in kleinen und mittleren Unternehmen, in denen kein oder nur eingeschränkter Zugang zu HPC-Supercomputern besteht.
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
The present dissertation answers the research question of efficient execution of optimization tasks in product development by using parallelization methods and by considering the existing resources within an organization. Based on analogy observations on the behavior of interdisciplinary teams, a method for efficient parallelization of evaluation processes of optimizations was developed and implemented prototypically into a framework for distributed optimization. Using the terms of concurrent optimization and simultaneous optimization, the methods of concurrent engineering and simultaneous engineering, which are commonly used for the parallelization of product development processes, are transferred to the application of optimization procedures.
The parallelization of evaluation processes is based on the decomposition of both the optimization model and the evaluation model. Single process elements of the evaluation process are allocated dynamically within the framework. Since the processing of these process elements depends only on the availability and the stability of required information and resources, high flexibility in the application and dynamic response to spontaneous changes in the resources can be ensured. To achieve a large distributed computing environment the units of a workstation cluster are used as resources. This allows the use of free computing capacities within the workstation cluster. Thus, this cluster becomes a very cost-efficient distributed environment. Within this environment the term “resource” also covers installed software and available licenses. Furthermore, the efficiency of an optimization can be increased by using priority-based processing of the process elements, since unnecessary idling of existing resources is thus avoided. The results of this thesis show that optimization methods are more effective and more flexible for determining the optimal priority values than heuristics. Thereby, the optimal priority values can by generated best by using the genetic algorithm NSGA-II. These priority values yield to the shortest processing time of the final product optimization.
The validation of the developed framework is based on three case studies, which represent different typical multidisciplinary optimization tasks of real product development projects. Thus, the execution of these case studies reflects an industrial environment where free workstations can become available as resources in a cluster during breaks, at night, or at weekends. The framework can be used in both homogeneous and heterogeneous workstation clusters, which can be implemented cost-efficiently from an arbitrary number of workstations connected by a network. Additional hardware is not required. Since data is transferred directly between these workstations, there is no need for a shared memory environment. Even during a running optimization available workstations can be added to the cluster or removed from it very quickly. Workstations of the cluster are available for interactive work at any time. Thus, interactive work is not affected by a running optimization. Therefore, the developed framework is suitable for applications in small and medium-sized enterprises which have no or only limited access to HPC supercomputers.
A survey of current continuous nonlinear multi-objective optimization (MOO) concepts and methods is presented. It consolidates and relates seemingly different terminology and methods. The methods are divided into three major categories: methods with a priori articulation of preferences, methods with a posteriori articulation of preferences, and methods with no articulation of preferences. Genetic algorithms are surveyed as well. Commentary is provided on three fronts, concerning the advantages and pitfalls of individual methods, the different classes of methods, and the field of MOO as a whole. The Characteristics of the most significant methods are summarized. Conclusions are drawn that reflect often-neglected ideas and applicability to engineering problems. It is found that no single approach is superior. Rather, the selection of a specific method depends on the type of information that is provided in the problem, the users preferences, the solution requirements, and the availability of software.
Wie entsteht ein neues Produkt oder eine neuartige Lösung? Immer schnellere Fahrzeugmodellwechsel verlangen einen immer besseren Produktentstehungsprozess. In einem gemeinsamen Forschungsprojekt zwischen Volkswagen und der Universität Magdeburg wurde hierzu ein Verfahren zur Anpassungskonstruktion entwickelt, welches auf der autogenetischen Konstruktionstheorie (AKT) aufsetzt. Der Lehrstuhl für Maschinenbauinformatik hat hierfür den Prototypen eines Programmsystems entwickelt, mit dem bereits verschiedene Aufgabenstellungen aus der Motorenentwicklung bearbeitet worden sind.
The main focus of this paper is to appraise Integrated Product Development (IPD) in industrial implementation and university teaching since the approach has been developed in the early 1980s by OLSSON and AN-DREASEN & HEEN. Integrated Product Development is still interpreted differently, in the range from development methodology to an idealised model of product development, as e.g. by ANDREASEN. But only the comprehensive description and understanding of the IPD approach as a development philosophy allows its application in all fields of product development. This philosophy contains four views: Planning and organisation, technology, methods & processes (especially with the inclusion of technical design), and human user (with the inclusion of ergonomics and working psychology). That is how it integrates all required factors for a successful product development, and, above all, fosters the finding of appropriate decisions with the right participants at the earliest possible time. This paper describes how the holistic product development philosophy of Integrated Product Development was implemented into the teaching at the Otto-von-Guericke-University Magdeburg (OvGU) and how this project oriented education is organized in integrative and interdisciplinary teams.
CAx-Systeme verwenden eine Fülle verschiedenster Modelle, die sich auf unterschiedliche Modellierungskonzepte stützen und
zu den wichtigsten Fundamenten von CAx-Systemen gehören. Den Grundlagen der Modellbildung kommt daher gerade im Zusammenhang
mit CAx-Systemen eine besonders große Bedeutung zu, weshalb in diesem Kapitel näher darauf eingegangen werden soll.
Aachen, Techn. Hochsch., Diss., 2003 (Nicht für den Austausch).
Autogenetischer Ansatz f'ur die Konstruktionstheorie. Beitrag zur vollst'andigen Beschrei-bung des Konstruktionsprozesses
- T Bercsey
- S Vajna
Bercsey, T. and Vajna, S. (1994) 'Autogenetischer Ansatz f'ur die Konstruktionstheorie. Beitrag
zur vollst'andigen Beschrei-bung des Konstruktionsprozesses', CAD-CAM Report 13(1994)2,
S. 66-71, und CAD-CAM Report, Vol. 13, No. 3, pp.98-105.
Die Entdeckung des Chaos (The Exploration of the Chaos)
- J Briggs
- D F Peat
Briggs, J. and Peat, D.F. (1990) Die Entdeckung des Chaos (The Exploration of the Chaos),
Carl Hanser Verlag, München.
Integrierte Produktentwicklung (Integrated Product Development)
- K Ehrlenspiel
Ehrlenspiel, K. (2007) Integrierte Produktentwicklung (Integrated Product Development),
Carl Hanser Verlag, München.
Development of an evolutionary design method
- K Kittel
- S Vajna
Kittel, K. and Vajna, S. (2009) 'Development of an evolutionary design method', Proceedings of
ICED 09, Design Methods and Tools, Part 2, Vol. 6, pp.147-156.
- G A Miller
- E Galanter
- K H Pribram
Miller, G.A., Galanter, E. and Pribram, K.H. (1991) Strategien des Handelns. Pläne und Strukturen
des Verhaltens (2. Auflage), Stuttgart Klett-Cotta.
Autogenetische konstruktionstheorie -ein beitrag für eine erweiterte konstruktionstheorie auf der basis evolutionärer algorithmen (Autogentic design theory -a contribution for a extended design theory based on evolutionary algorithms)', Dissertation, Otto-von-Guericke
- B Wegner
Wegner, B. (1999) 'Autogenetische konstruktionstheorie -ein beitrag für eine erweiterte
konstruktionstheorie auf der basis evolutionärer algorithmen (Autogentic design
theory -a contribution for a extended design theory based on evolutionary algorithms)',
Dissertation, Otto-von-Guericke-Universität Magdeburg, Magdeburg.