Application of artificial intelligence techniques in seismology and engineering seismology

Abstract

In the field of risk evaluation and seismic hazard assessment, it is necessary to codify a great quantity of aspects of the so called knowledge and to supply an intelligent support for the not-well-defined problems (data uncertainty, lack of rigorous solution algorithms). The main feature of an expert system is to emulate effectively the behaviour of a human expert in a particular and defined field, enabling the final user to improve its decisional process and giving access to him to a knowledge base otherwise not clearly codified. From these general considerations the intention came to develop the prototype CZAR (Classificatore Zone A Rischio) that is an expert system reproducing the Italian seismic classification based on the definition of Seismic Hazard given by Progetto Finalizzato Geodinamica (PFG) of the Consiglio Nazionale delle Ricerche (CNR). The expert system built up on the commercial shell Nexpert Object is working on a personal computer through a graphic interface developed with the Graphical User Interface (GUI) of Window 3.0. This user friendly interface makes possible the choice of different procedures to estimate the hazard parameters and also allows the activation of the classification inferential process. The influence of different assumptions and strategies has been evaluated by a mathematical algorithm suggested by the general structure of the Bayes’ theorem. In this paper the prototype of the expert system has been applied to the data relating to Toscana region (central Italy) and the interactive evaluation of the maps furnishes a relative measure for discrepancies on seismic classification in the 2nd seismic category.