DECISION-MAKING IN MANUFACTURING: AN INFORMATION PROCESSING APPROACH

The broad objective of this research is to employ concepts from optimization theory, artificial intelligence, and computer architecture to the analysis of organizational problem-solving. Abstract organizations can be characterized as consisting of information processors or resources that are interconnected through a network. A limited commonality of interest exists between the processors. The primary task is to use the resources to carry out some tasks or achieve some goals. The purpose of this research is to model the manufacturing component of an organization and to explore various scheduling problems associated with this approach. The computer-based information system that models such a problem-solving system consists of three components: a knowledge base in which both general and domain-dependent knowledge will be represented, a natural language interface through which the user can communicate with the system, and a problem processor that carries out a combination of data retrieval, computation and deduction in order to respond to queries posed by the user. An axiomatic framework based on equational logic is proposed to model the computer-based information system. The use of this framework to represent environmental and linguistic knowledge as well as perform the language processing and problem-solving functions in a uniform manner is demonstrated. As an illustration of the functioning of the problem-solving component, a framework for the on-line scheduling of jobs as they enter the system is introduced. A heuristic solution based on artificial intelligence techniques is proposed for the off-line or static scheduling problem. The solution is then extended to a more dynamic environment where jobs continually arrive over time. Finally, a model for the decentralized control of a general multilevel manufacturing system is proposed. Future research directions are suggested