Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 74
PROCEEDINGS OF THE SIXTH INTERNATIONAL CONFERENCE ON THE APPLICATION OF ARTIFICIAL INTELLIGENCE TO CIVIL AND STRUCTURAL ENGINEERING
Edited by: B.H.V. Topping and B. Kumar
Paper 3

Search-Control Knowledge for the Interoperability Problem between Conceptual and Preliminary Structural Design

M. Eisfeld and R.J. Scherer

Institute of Applied Computer Science in Civil Engineering, Dresden University of Technology, Dresden, Germany

Full Bibliographic Reference for this paper
M. Eisfeld, R.J. Scherer, "Search-Control Knowledge for the Interoperability Problem between Conceptual and Preliminary Structural Design", in B.H.V. Topping, B. Kumar, (Editors), "Proceedings of the Sixth International Conference on the Application of Artificial Intelligence to Civil and Structural Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 3, 2001. doi:10.4203/ccp.74.3
Keywords: conceptual structural design, preliminary structural design, structural model, description logic, hierarchical planning, declarative control knowledge, reasoning tasks.

Summary
This paper discusses a computational framework for the conceptual and preliminary structural design (CPSD) process. We establish the computational framework on the knowledge level, because we consider the designer as a symbolic processing system. Therefore we can functionally define the common sense knowledge types and the task structure of the CPSD process from practitioners. The framework provides control facilities by suitable methods for knowledge representation, where we have obtained the appropriate methods by analysed design studies.

We distinguish between procedural, declarative, and task-implementation knowledge, which contribute to the overall search-control in the design space at the knowledge level[1]. First, we give an account of the "ill-structuredness" of the CPSD problem in terms of applicable methods from Artificial Intelligence as motivated in[2], by which we later represent the different knowledge types. Second, we specify the constituents from the F-B-S design model for our structural design domain. Third, we introduce a formalism based on first order predicate situation calculus for the representation of common sense procedural knowledge This knowledge type has been successfully applied by hierarchical refinement in the domain of conceptual and preliminary structural design[3], because the method shows mainly advantages for the implementation of procedural knowledge in design assistant systems due to its flexibility. Fourth, we define a structural model in description logic, that evolves on the domain ontology during the design process. The structural model is defined by declarative knowledge about structures, its components, their relations and functions. We consider two abstraction levels for the structural model, namely the global functional one and the component one. The structural model evolves on the global level during the conceptual design and on the component level during the preliminary design. We employ these abstraction levels of the structural model for reasoning tasks in the design domain. The reasoning tasks consequently represent task-implementation knowledge.

CPSD is a complex and heterogeneous process. The distinction between the conceptual and preliminary design stage in reality has yet not been formalized in design assistant systems. This clear distinction is featured by the use of different abstraction levels of the structural model in the two sub-problem spaces. The structural model enables reasoning tasks at the transition from the conceptual into the preliminary design stage and at the boundary of the preliminary design stage. They represent control-knowledge about the progression into the deeper preliminary design space and about the selection of gross-sectional member parameters. The tasks are essential to verify the so far developed structural concept, to control the amount of applicable design operators, to decrease the complexity of the overall design task and to select the required gross-sectional parameters according to the behaviour of the structural system. They are encapsulated into the action parts of design operators in the design domain and take the so far developed taxonomic structure of the load bearing system into account.

Our chosen approach places emphasis on the CPSD process as a task, which is controlled by reasoning about the evolving structural model in different design situations[4]. Without the clear distinction of the conceptual and preliminary design phase, which employ different abstraction of the evolving structural model, the complex design process can hardly be controlled for design assistant programs.

References
1
Newell, A., "The Knowledge Level", Artificial Intelligence, 18:87-127, 1982. doi:10.1016/0004-3702(82)90012-1
2
Simon, H.A., "Models of Discovery: And Other Topics in the Methods of Science", D. Reidel Pub. Co., Dordrecht, 1977.
3
Hauser, M. and Scherer R. J., "Application of Intelligent CAD Paradigms to Preliminary Structural Design", Artificial Intelligence in Engineering 11(3): 217 � 229, 1997. doi:10.1016/S0954-1810(96)00042-8
4
Gero, S.J., "Conceptual Design as a Sequence of Situated Acts", Lecture Notes in Artificial Intelligence, No. 1454, Springer, Berlin, 1998. doi:10.1007/BFb0030450

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description