| ID:10001 |
| Type of Publication: Journal Articles |
| Authors: Alexander Galperin, Y.Kimhi, E.Nissan, H.T.Siegelmann |
| Title: FUELCON's heuristics, their rationale, and their representations |
| Name of the Journal: New Rev. Appl. Expert Syst. (UK) |
| Year: 1998 |
| Volume: 4 |
| Pages: 163 - 76 |
| Abstract: The FUELCON expert system makes computations for reactor core reload design more transparent in several ways. First, the heuristics in its ruleset reflect a high level abstraction of the human expert's operational knowledge; a few mandatory rules are mandated by safety considerations, whereas a few other rules embody a particular strategy of search. Second, FUELCON does not just look for one satisfactory solution, it generates a multitude of candidate configurations, resulting in the search policy embodied in the ruleset. Once these solutions are visualized as dots in the bidimensional plane of cycle length and power peak, not only does this enable the user to pick the few most promising solutions for further consideration, but the domain expert, qua user, of FUELCON can revise the ruleset to emphasize a trend he spotted in the visualization and would like to explore. Ruleset revisions enable “zooming” on subregions of the search space. An ergonomic methodology of how to best use FUELCON eventually came into existence |
| Keywords: data visualisation;expert systems;heuristic programming;knowledge representation;nuclear engineering computing;nuclear power stations;power engineering computing; , |
| Last Updated: 22/11/2006 00:00:00 |
