The mechanical properties of rock mass are so complicated that some scholars believe that rock mechanics and rock engineering are perhaps as much art as science. Therefore, expertise may play an important role in these fields in which the Expert Systems from Artificial Intelligence are developed rapidly update as a result. To elicit knowledge from experts is the bottle-neck which brings some difficulties for building expert systems. A case-based reasoning system appeared lately may overcome this obstacle. In this paper, we present a case-based reasoning system for tunnel support design to demonstrate the merits of this method.
Les proprietes mecaniques de la masse du roc sont tellement compliquees que certains savants croient que la mecanique et la technique de I' ingenieur du roc seraient autant d' art que la science. Donc I' expertise peut jouer un rôle bien important dans cettes regions, auxquelles systèmes d' expert tires de l' intelligence artificlelle etaient developpes rapidement par suite de moderniser. Mais obtenir connaissance d' expert est le goulot de bouteille qui eprouvet quelques dlfficultes à construire le système d' expert. Un système de raisonnement base au cas apparu par la suite peut vaincre cet obstacle, et une sorte du système pour dessein du soutien de tunnel etait presentee dans ee papier pour demontrer les merites de la methode.
Die mechanischen Elgenschafte der Rockenmasse sind so verwickelt, daβ jemand mitten aus Gelehrten glaubt, die Rockenmechanik und das Rockeningenieurwessen vielleich so viel die Kunst wie die Wissenschaft sind. Daher die Geschlcklichkelt mag eine wichtige Rolle in diesem Feldern spielen, worin das Kundigsystem aus der Intelligenz entwickelnt schnell als Ergebnis der Neuzeitlichkeit ist. Es ist der Flaschenhals, das Wissen der Sachverstandiger herauszulocken, was etwas Schwierigkeiten fuer Kundigsystengebaude bringt. Das Verweisfuehrungsystem gegruendet auf Umstanden erscheinet kuerzlich mag das Hindernis ueberwaltigen. Dieses Papier einfuehrt ein solches System fuer Konstruktion der Tunnelunterstuetzung, die Verdienste der Methode zu Eigen.
Expert systems are developed rapidly in rock mechanics and rock engineering (Zhang et al. 1988 &. 1993) in recent years. Building an expert system with rule-based knowledge is time consuming because rule extraction from human experts Is a labor- intensive work. This is why some scholars consider that the knowledge acquisition is a bottle-neck for building an expert system. The case-based reasoning (CBR) was first proposed by R. Schank and C. Riesbeck (1989). CBR Industry applications are being built and fielded at Lockheed, GTE, DEC, Boeing and etc.. Case-based reasoning is rapidly emerging from AI because it can use past experiences (cases) to solve current problems without the trouble of knowledge acquisition. Case-based reasoning systems operate in a different way from expert systems. Given an input specification, a case-based system will search its case memory for an existing case that matches the input specification. It is Impossible that a case searched from the case memory entirely matches the input specification in practice, but one or some cases which are similar to the input situation may be found absolutely. The case-based system must then find and modify some portions of the retrieved cases that do not meet the input specifications even most portions of them are similar. This is called case adaption. The result of case adaption is a completed solution, but it also generates a new case, after putting it into practice, that can be automatically added to the system's case memory for future use. In general, case-based systems are easier to build and maintain because the knowledge engineering task is reduced to the simple problem of defining terms and collecting preclassified cases from experts or the case studies.