Knowledge engineering of an intelligent system for selection of remediation methods at petroleum contaminated sites

A rule-based expert system for the management of petroleum contaminated sites is described which was developed using a variety of methodologies such as Inferential Modelling Technique (IMT) for knowledge analysis, decision trees and object-oriented methodology for knowledge representation. To enhance the system's flexibility, a user interface and an engineer interface were constructed for capturing different types of input. The system was implemented in real time with the real time expert system G2. An important feature of the system is its capability of dealing with imprecision in the inputs on the site conditions and contamination levels. The developed system contributes to the selection task by allowing for a more comprehensive and realistic consideration of the many factors related to the remediation selection decision. The results from case studies indicate that the expert system can aid the environmental engineer identify the most optimal alternatives for site remediation and thereby help industry reduce costs of site remediation practices. The system's inputs and outputs are directly useful for further modelling studies on remediation processes and contaminant transport in soil and groundwater. 10 refs., 6 figs