REPRESENTATIONAL ISSUES IN A KNOWLEDGE-BASED JOB SHOP MANAGEMENT SYSTEM

WE PROPOSE TO DESIGN, IMPLEMENT AND TEST A KNOWLEDGE-BASED SYSTEM FOR THE DYNAMIC SCHEDULING OF JOB SHOPS. THE TASK OF SCHEDULING A JOB SHOP IS INHERENTLY DIFFICULT; INCLUDED ARE CONSIDERATIONS SUCH AS FLEXIBILITY, PROFITABILITY, TRACEABILITY, QUALITY ASSURANCE AND ERROR DETECTION AND CORRECTION. THE TWO KEY COMPONENTS OF THE PROPOSED SYSTEM THAT WE PLAN TO INVESTIGATE IN PHASE I ARE: (1) SELFCORRECTING INFERENCE MECHANISM WHICH PERFORMS PLANNING BASED ON KNOWLEDGE OF CONSTRAINTS IN THE DOMAIN AND DOES RE-PLANNING DURING PLAN EXECUTION ON THE BASIS OF NEW UNFORSEEN INFORMATION THAT MAY ARISE; AND (2) A GENERIC REPRESENTATION OF KNOWLEDGE THAT ENABLES MULTIPLE DIFFERENT REPRESENTATIONS OF COMPLEX OBJECTS (E.G., PARTS, MATERIALS, MACHINES), IN ORDER TO ALLOW THE TRACKING OF OBJECTS THROUGH THE JOB SHOP DURING THE EXECUTION OF A JOB PLAN.