An expert system for concrete diagnosis

Abstract

Concrete is a material liable to deterioration atributable to either external or internal influences. Diagnosis a defect can be established through identification and determination of its dominant cause. In this study, a knowledge-based system, "CONCEXS", is developed to diagnosing several concrete defects frequently occurring in concrete slabs, beams, and columns. This study addresses the salient notions of concrete diagnosis and particularly handles some deterioration problems for which relevant testing techniques are summarized. The knowledge base is acquired from literature expertise and domain exerts then employed to analyze existing defect cases in Saudi Arabia. Defects' attributes are heuristically formulated into decision trees that are suitable for encoding as production rules. CONCEXS-user interface is efficiently interactive by means of call up menus thus guiding the user through the diagnostic process. The system utilizes simple linguistic values which exhibit friendly inteaction during a consultation session. Verification and validation of CONCEXS were executed, and proved its capability of deducing diagnostic decisions satisfactorily.