An investigation of the applicability of expert systems to job shop scheduling

Abstract

Although the job shop scheduling (JSS) problem is of eminent practical importance and has received considerable attention by both industry and academia, because of its complexity, it still remains an enigma. Artificial Intelligence holds the potential for providing solutions to complex problems, such as the JSS problem. However, due to the magnitude of the JSS problem, search techniques are not computationally feasible. Expert Systems, though, have been successfully applied to problems of this type. The development of an expert system, however, implies the availability of an expert. Regrettably, such experts are not readily available in the JSS environment. The approach presented in this paper involves the use of computer simulation models of a job shop to train subjects so that they are capable of effective scheduling, and then extracting the knowledge of these “experts” to develop an expert system.