EXPERT SYSTEMS IN CIVIL ENGINEERING DATA ANLYSIS

Due to the proliferation of microcomputer statistical software the need to incorporate more intelligence in such systems has become paramount. Expert systems have the potential to substantially aid data analysis by incorporating guidance, explanation of output, information retrieval and diagnostic checking into statistical software systems. This type of software can be of significant benefit to engineers or analysts who have limited statistical training or no direct access to a statistician. However, there are several areas of concern in applying expert systems to data analysis. These relate to problem definition and the selection of the most appropriate analysis procedure to use. These obstacles have forced the development of highly specialized statistical expert systems which seem to offer the most effective method of assisting data analysis. "transtat" is a probability density analysis package designed to perform distributional analysis of engineering systems. It contains several small expert systems to aid this task. A small model guidance module has been developed to assist model selection. A 'short list' of model families that have a high potential of representing sample data can be produced. Other applications of expert systems in "transtat" include outlier detection and the inclusion of concise and jargon free explanations of output from the goodness of fit tests. This paper describes many of these features in a civil engineering data analysis context (a).

  • Availability:
  • Corporate Authors:

    Monash University

    Department of Civil Engineering
    Clayton, Victoria  Australia  3800
  • Authors:
    • Thompson, Russell G
  • Publication Date: 1988-6

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 00483182
  • Record Type: Publication
  • Source Agency: Transport and Road Research Laboratory (TRRL)
  • Files: ITRD, TRIS
  • Created Date: May 31 1989 12:00AM