The development of an expert system for finding fragility curves of building structural systems in the preliminary design stage

This research is a starting point for the development of an expert system for determining seismic fragility curves of structural systems in a nuclear power plant or conventional building at the preliminary design stage. The resulting system will assist an engineer with moderate engineering background and limited reliability knowledge to analyze the failure functions of building structures. It simulates the performance of human experts in identifying the potential failure modes and their variabilities for a structure of interest. Then, by combining the stress results of crude analyses, the method of linear statistical modelling, the current conceptions of seismic fragility evaluation and damage evaluation, and the intuition and experience of human experts, the seismic fragility functions of the structure will be estimated.; The induction technique will be used for knowledge acquisition. Thus, the characteristics of structural system performance during earthquake excitations will be described by attributes and their values. The rules for the selection of the components for seismic fragility evaluations may be deduced from the results of expert diagnosis of a pattern space configured from the attributes. Because the set of attributes and their values is generally incomplete, the deduced rules often contain uncertainties and imprecisions. Also, the local responses estimated by crude analyses, approximations of analytical modelling, empirical formulas and tricks will result in the other sources of uncertainties. To support this inexact reasoning inference process, fuzzy set theory will be employed.; A procedure to combine the implementation techniques of a knowledge base expert system and traditional programming has been developed in order to build an expert system for solving a semi-ill-formed problem as seismic fragility evaluation of building structural systems in the preliminary design stage. Based on this implementation technique, the major sequential order associated with the reduced subproblems was maintained; however, the expertise and facts related to each subproblem has been written in terms of a knowledge base expert system. This design strategy decomposes this complicated problem into a series simpler subproblems, fires the associated rule sets, fact bases, empirical formulas and tricks to solve the subproblems, and combines the results of these subproblems to estimate the solution. Hence, the domain of search space and computer time will significantly be reduced.