AN EXPERT SYSTEM TO DISCOVER CAUSE OF FAILURE FOR IBM MICROCOMPUTER PERIPHERALS INTERFACE

A knowledge-based system to discover cause of failure for the IBM microcomputer peripherals interface has been built using a backward-chaining inferencing expert system framework. A detailed technical reference manuals are currently used for peripherals interfacing maintenace and failure diag-nosis. Task complexity limits this method. A set of real-time self-test rou-tines, database, explanation facility and interactive menus have been built to compensate for the inadequate man-machine interfaces to this system. The resulting diagnosis system, we call it MUES (Menoufia University Expert System) incorporates: (i) databases to store data, control and status regis-ters contents, I/O ports configuration modes and communication format, (ii) interactive menus, and (iii) an intelligent rule-based system which takes its evidence in real-time via a set of assembly modules and provides advice on the cause of failure to the operator. MUES written in Turbo Prolog and running on an IBM XT microcomputer, uses if-then production rules to encode the knowledge and is based on the backward-chaining inference engine that allows rules to look very much like English. The interaction between the operator and the knowledge-based diagnosis system has been reduced to a small number of questions requiring yes/no answers. The explanation facility requires how/why questions. The remaining input to the diagnosis system is provided by status.evidence (logic values low/high) as a result of executing a set of self test routines written in assembly language.