caa a22 4500 469072784 CHVBK 20180323132926.0 cr unu---uuuuu 170328e19921201xx s 000 0 eng 10.1007/BF02032463 doi (NATIONALLICENCE)springer-10.1007/BF02032463 Peters Thomas Department of Computer Science and Engineering, University of Connecticut, 06269-3155, Storrs, CT, USA aut Mechanical design heuristics to reduce the combinatorial complexity for feature recognition [Elektronische Daten] [Thomas Peters] Feature recognition can be characterized as enhancing the geometric database representation of a mechanical artifact to include some design intent. A number of feature recognition systems have successfully demonstrated the ability to capture such design knowledge. To determine if any such system could "scale-up” to large industrial part databases, it is appropriate to analyze the computational complexity of the recognition system-an issue which appears to have generally been ignored by the feature recognition community. A naive combinatorial analysis, generally applicable over all design domains, suggests that the combinatorial complexity would be exponential, hence intractable within most modern computing environments. Such a theoretical mathematical analysis ignores critical domain-specific engineering knowledge. It is argued that this theoretical potential for combinatorial explosion can be addressed by combining domain-specific mechanical engineering heuristics with a detailed combinatorial analysis. An example of such a heuristically guided combinatorial analysis will be presented for the domain of sheet-metal parts. This example presents combinatorial upper bounds which indicate that feature recognition upon large sheet-metal databases will be computationally tractable. In order to extend the analysis to other domains, it may be necessary to develop other domain-specific heuristics. However, the example presented should serve as a guide to incorporating design heuristics with combinatorial analysis. Springer-Verlag New York Inc., 1993 Research in Engineering Design Springer-Verlag 4/4(1992-12-01), 195-201 0934-9839 4:4<195 1992 4 163 https://doi.org/10.1007/BF02032463 text/html Onlinezugriff via DOI 1 research-article jats NATIONALLICENCE 856 40 https://doi.org/10.1007/BF02032463 text/html Onlinezugriff via DOI NATIONALLICENCE 100 1- Peters Thomas Department of Computer Science and Engineering, University of Connecticut, 06269-3155, Storrs, CT, USA aut NATIONALLICENCE 773 0- Research in Engineering Design Springer-Verlag 4/4(1992-12-01), 195-201 0934-9839 4:4<195 1992 4 163 Metadata rights reserved Springer special CC-BY-NC licence nationallicence BK010053 XK010053 XK010000 NATIONALLICENCE NATIONALLICENCE NL-springer