Some of the factors which impinge upon rock excavation design, and which might affect the stability and safety of such excavations, are outlined. Expert systems, neural networks and other forms of artificial intelligence are described briefly, and applications of these techniques to various problems associated with rock engineering are reviewed. Exploratory expert systems have been developed at CSIRO to guide a practising engineer in the use of computational stress analysis in excavation design; these are discussed. Finally, some suggestions are made regarding possible future advances in rock excavation safety through the wider use of expert systems and neural networks.
Quelques facteurs intervenant dans le projet d'excavation de roches, et qui pourront done affecter la stabilite de ces excavations, sont soulignes. Des systèmes experts, des reseaux neuraux et d'autres types d'intelligence artificielle sont succinctement decrits, et des applications de ces techniques aux plusieurs problèmes lies au genie des roches sont revisees, Des systèmes expert d'interrogation ont ete developpes à CSIRO pour guider les ingenieurs dans l'utilization de l'analyse informatique de contraintes dans le projet d'excavation; ces systèmes sont objet de discussion. Finalement, quelques suggestions sont presentees en ce qui concerne les futurs developements possibles par rapport à la securite de I'excavation de roches moyennant une utilization plus intensive de systèmes experts et de reseaux neuraux.
Faktoren die beim Bau unrertagiger Hohlraume eine Rolle spielen und die Stabilitat und Sicherheit solcher Hohlraume beeinflussen werden dargestellt. Expertensysteme, neurale Netzwerke und andere Formen kuenstlicher Intelligenz werden kurz beschrieben und es wird ein Uberblick gegeben ueber Anwendungen dieser Methoden im Zusammenhang mit verschiedenen felsmechanischen Fragestellungen. Testvarianten von Expertensystemen wurden vom CSIRO entwickelt mit dern Ziel den Praxis-ingenieur bei numerischen Spannungsberechnungen und Hohlraum Bemessungen zu unterstuetzen. Schliesslich werden Vorschlage gemacht im Hinblick auf zukuenftige Fortschritte im Zusammenhang mit Hohlraum Sicherheit durch verstarkte Verwendung von Expertensystemen und neuralen Netzwerken.
The theme for Workshop W4 at EUROCK '93 was "Expert systems as a tool for safety evaluation". The workshop did not assume any prior experience with expert systems by the participants. Rather, it was intended to provide an introductory overview of what expert systems and other forms of artificial intelligence are and can do; to outline the classes of problem which they might help with; and to discuss previous and possible future applications of these techniques to problems related to safety in rock engineering. This paper is based on a presentation by the author, as leader of the discussion panel for Workshop W4. Its role is to provide an introduction to the topic, by means of a discussion of the applicability of expert systems as a guide to rock excavation design. Rock excavation safety depends upon the stability, predictable instability, or engineered support of the rock mass surrounding an excavation. In designing an excavation, an engineer seeks an optimum balance between safety and economy, endeavouring to minimise costs associated with, and maximise benefits gained from, creation of the excavation. The balance point will be different for "permanent" and "temporary" excavations (e.g. for mine shafts compared with stopes). It may also differ from mining to civil engineering excavations, with the latter generally needing to be safe for longer periods of time. Safety and economy may also be closely intertwined. In underground mining, instability of roof strata or a stope wall may endanger life. It may also result in dilution of are by the addition of failed (waste) rock to the are stream extracted from the mine, increasing processing costs for no additional product.
