«A Computer Application to Study Engineering Projects at the Early Stages of Development M. H. Gedig M.A.Sc Structural Engineer AGRA Coast Inc. 1515 ...»
Qualitative and semi-quantitative methods have distinct advantages over other approaches for dealing with uncertainty in engineering systems, such as probability methods, Monte Carlo simulation, hill-climbing techniques and expert systems. Numerical simulation techniques based on probability, such as Monte Carlo methods, and hill-climbing techniques, which seek the maximum and minimum values of each quantity, are not sound inferential techniques. and thus may exclude legitimate solutions. They generally return a subset of the true range, and thus arbitrarily exclude legitimate possibilities. The techniques discussed in this paper are logically sound. The interval analysis methods employed are able to guarantee bounds on the correct solution.
Expert systems represent another approach to coping with uncertainty in engineering practice.
These systems have been applied successfully in engineering domains, partly because they are able to reason with valuable experiential and heuristic knowledge. Most expert systems suffer, however, because of an inability to reason using fundamental domain knowledge. On the other hand, qualitative reasoning uses a detailed domain model, an explicit representation of the first principles of the domain that underlie heuristic knowledge.
Purely qualitative reasoning methods developed in artificial intelligence research were applied to a practical problem in the domain of structural engineering. It was found that the conclusions that may be drawn from such methods could be quite powerful when the number of variables in the problem is limited. Unfortunately the conclusions that may be drawn become rapidly weaker as the number of unknowns in the problem is increased. Although a limited number of problems were tested, the findings are not out of line with previous research. A number of attempts have been made by artificial intelligence researchers to resolve the ambiguity of predictions made using qualitative analysis. The approach used in this research was to augment qualitative analysis with partial quantitative information. Semi-quantitative techniques show considerable promise in the practice of engineering for their ability to model design abstraction while providing bounds for the behaviour of physical systems.
6.1 Further Research In order to fully evaluate the techniques presented in this paper, the techniques must be used in real engineering applications. The goal of developing the QES software was to provide an accessible tool, which could be used, for the analysis of conceptual and preliminary engineering designs. Further research is required to ascertain which types of reasoning, qualitative or quantitative, are best suited various types of engineering problems.
1. Davis, E. (1987) Constraint propagation with interval labels. Artificial Intelligence 32:281Forbus, K.D. (1984) Qualitative process theory. Artificial Intelligence 24:85-168.
3. Freuder, E.C. (1978) Synthesizing constraint expressions. Communications of the ACM, 21:958-966.
4. Gedig, M.H. (1995) A Framework for Qualitative and Semi-Quantitative Analysis in Engineering Design and Evaluation. Masters Thesis, University of British Columbia.
5. Knuth, D.E. (1968) The Art of Computer Programming. Addison-Wesley, Reading, MA.
6. Kuipers, B.J. (1994) Qualitative Reasoning: Modeling and Simulation with Incomplete Knowledge. MIT Press, Cambridge, MA.
7. Lhomme, O. (1993) Consistency techniques for numeric CSPs. In Proc. Thirteenth Joint Conference on Artificial Intelligence, IJCAII 1993, Morgan Kaufmann, San Mateo, CA.
8. Mackworth, A.K. (1977) Consistency in networks of relations. Artificial Intelligence 8:99McDermott, D.V. and Davis, E. (1984) Planning routes through uncertain territory.
Artificial Intelligence 22:107-156.
10. Moore, R.E. (1966) Interval Analysis. Prentice-Hall, Inc., Englewood Cliffs, NJ.
11. Simmons, R. (1986) Commonsense arithmetic reasoning. Proc. Fifth National Conference on Artificial Intelligence 118-124.