A Fuzzy Expert System: Applications to Geology

J. H. Fang

Expert systems are potentially valuable tools for scientists who are in so-called inexact sciences, such as geology. This is because expert-system technology deemphasizes the number-crunching aspect of computers; instead, symbol manipulation is the kernel of the expert-system software. Most geologic data are imprecise and often incomplete; yet, geologists must make decisions under uncertainty based on approximate reasoning. To deal with those situations, Aristotelian logic of true/false employed in most expert-system shells is inadequate; fuzzy logic must be incorporated in the inference engine as well as knowledge base of the expert system. Furthermore, many geological problems cannot be solved by a deductive chain of reasoning, but the solutions have to rely on a collec ion of related facts in an inductive fashion. An expert system shell which can accomplish all these is FLOPS--Fuzzy Logic Production System. The system employs fuzzy concepts throughout: production rules use fuzzy logic; data types include fuzzy numbers; and fuzzy predicates are provided for fuzzy comparison of numbers. It handles inductive reasoning by a production system which fires rules in parallel rather than sequentially, using weakly monotonic logic to solve memory conflicts. This paper is illustrated by examples from qualitative x-ray phase analysis and determination of clastic depositional environments.

AAPG Search and Discovery Article #91038©1987 AAPG Annual Convention, Los Angeles, California, June 7-10, 1987.