ESCAR-AN EXPERT SYSTEM FOR CRACK DIAGNOSIS AND REPAIR IN BUILDING

Abstract

Since World War II, computer scientists have tried to develop the techniques that would allow computers to act more like humans called as Artificial Intelligence (AI). A collection of AI techniques that enables computer to assist people in analysing problems and making decisions, called Knowledge Based Expert Systems, has recently proved its value and today numerous commercial applications are underway. The expert system is an AI software incorporating a knowledge base and an inferencing system. The expert system attempts to duplicate the functi'ons of an expert in some field of expertise. The program acts as an intelligent consultant or adviser in the domain of interest, capturing the knowledge of one or more experts. Non experts can then tap the expert system to answer the questions and make decisions in the domain. The purpose of expert system is not to replace the experts, but to make their knowledge and experience, more widely available. For many real world problems the algorithm can not be formulated and Expert system provides the best solution for such problem. This dissertation addresses the problem of design, development and implementation of a rule based expert system for crack Diagnosis and Repair suggestions in building and named as ESCAR. The proposed system has about 140 rules and has been implemented in PC based LISP dialect GCLISP. The proposed system is developed in three modules, namely, 1]Consultation module - which communicates with the user through user interface to accept the symptoms and allied information and conveys the conclusion in the form of diagnosis and remedial measures for cracks. 21Assistance module - which helps the user to chalk out the repair procedure and provides information about available materials and techniques. 3]Rule Acquisition module - which provides the facility for modifying and updating the knowledge base of the system. A password is required to have a session with this module to save the knowledge base from unauthorised access. The knowledge base of the proposed system is in the form of production rules, drafted in IF-THEN format and developed through case studies, documentations available and through active participation and disscussions with experts in the domain. It must be realised that the expert system can only mimic the human intelligence a limited way and hence can never replace the human experts. Also the development of an expert system is extremely tedious as compared to conventional software.