Back to search

National Technical University of Athens (NTUA)

Εκτίμηση της υποστήριξης υπόγειων έργων με εφαρμογή τεχνικών τεχνητής νοημοσύνης

Abstract

dc:description

The application of Artificial Intelligence computer methods is investigated, in order to develop computational software that predicts successfully design parameters for the construction of underground sites. Two basic steps are followed, knowledge acquisition and suitable representation for use in a computer program. Knowledge is acquired through bibliographic research from expert recommendations and a search for previously used artificial intelligence applications. Various expert recommendations were put together in unified tables. Measured data from previously constructed underground works were gathered in a database. It was found by statistical analysis that the support measures of the underground openings stored in the database are related to the input data collected and therefore these measures can be estimated with the use of artificial intelligence. The representation of gathered knowledge and experience included the selection of the most important input variables used by the experts, the unification of common expert rules used to define a recommendation and the quantification of support measures. Artificial Intelligence (AI) methods used are expert systems, fuzzy logic and neural networks. There exist two main categories of inference that reach prediction. Rule based inference used by expert systems and fuzzy logic systems and data driven inference used by neural networks and classic statistical methods. Support measures are predicted either by two-input systems or by systems that use more input variables. Two-input systems predict support using the span of the opening and the rock mass quality index Q, similarly to the Q system. More input variables are used by systems that do not need an intermediate rock quality index. All AI systems’ predictions are compared to measured data and Q system predictions. It was found that the AI systems, that are using database records for their predictions, present an ability to pick up hidden relations among data and generalize more or less successfully. Rule based AI systems have a similar performance, suggesting that the rules of the experts are specific and correct. The two-input systems predict better than the Q system but their correlation is not too high. The many-input systems display higher performance indices than those of the two input systems. This suggests that more variables should be considered in underground site design bypassing the determination of a rock mass quality index.

Degree

thesis:*
Grantor dc:publisher
National Technical University of Athens (NTUA)
Year dc:date
2007

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Tzamos, Stergios
  • Τζάμος, Στέργιος

Subjects

dc:subject × 15

Rights

Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/16286

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Tzamos, Stergios; Τζάμος, Στέργιος. Εκτίμηση της υποστήριξης υπόγειων έργων με εφαρμογή τεχνικών τεχνητής νοημοσύνης. National Technical University of Athens (NTUA), 2007. http://hdl.handle.net/10442/hedi/16286