DESPLATE: An expert system for abnormal shape diagnosis in the plate mill

Abstract

The authors report the development of a diagnostic expert system, DESPLATE, to assist less experienced operators to diagnose the off-square, camber, and taper shapes of plates. The system also incorporates graphics and animation routines written in Turbo-Pascal to provide a more user-friendly interface. DESPLATE has been found by experienced operators to be capable of providing systematic guidance in searching for faults. The authors first provide a brief overview of the production process in the plate mill and describe the problems that motivate the development of DESPLATE. The goal of DESPLATE is to locate possible causes for abnormal shapes, such as electrical failures, mechanical breakdowns, wear, operational errors, or prerolling conditions. They describe the knowledge base architecture and user interface and examine the control structure of the system. The authors address some of the issues that arose during the development of DESPLATE, including the use of graphics for the interface, forward and backward chaining techniques, and knowledge acquisition methodologies from multiple experts from different disciplinary backgrounds.