Fuzzy expert systems in civil engineering

This research recognizes the potential value of Expert Systems in future civil engineering practice. However, in view of the inherent imprecision in civil engineering knowledge, this thesis considers the specific development of Fuzzy Expert Systems. In building Fuzzy Expert Systems the ability to model uncertainty and imprecision is recognized as the major pre-requisite. The thesis has three major objectives: 1. To emphasize the role that Expert Systems can play in civil engineering, identify areas of potential application within civil engineering and develop a few example applications. 2. To present a coherent treatment of the various inter- disciplinary concepts from statistical inference, information entropy, classical and non-classical logic. cognitive sciences and multi-attribute decision-making, which are pertinent to the development of Expert Systems. The aim is to develop a theoretical background for Expert Systems from a civil engineering point of view. 3. To undertake a detailed study of aspects of building Fuzzy Expert Systems which require a treatment within the context of civil engineering and cannot therefore be considered the sole responsibility of the Information Technologist. In this respect models of inference and implementation formalisms are considered in detail. An appraisal of existing models of uncertain and inexact inference with reference to typical civil engineering problems is followed by a proposal for a new model of inexact inference. The study also serves the purpose of illustrating the method of fuzzification of seismic design parameters. The seismic design examples have been selected in view of the long-term objective of presenting the seismic code as a Fuzzy Expert System. In investigating implementation formalisms for Fuzzy Expert Systems, the thesis demonstrates that logic programming whilst fundamentally classical, is capable of realizing the non-classical inference systems. However implementation of most inference models requires the ability for numerical computations, for which logic programming has traditionally been considered unsuitable. This aspect is investigated and the possibility of using a mixed language interface is examined. Finally, a logic programming language micro-PROLOG and the Expert System shell APES are used to develop two examples of civil engineering Expert Systems, GEOTECH and EXP-ATC.