Date of Award

1993

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

First Advisor

L. Ken Keys

Abstract

The explosion of computer applications into the world of manufacturing along functional lines has produced the often mentioned "islands of automation." Although many issues and problems are involved in interfacing and integrating the databases that serve these applications, we can extract valuable data from these independent systems to provide important information to decision makers. This research resulted in the development of MIMIR (Multiple Integrated Manufacturing Information Resources), a decision support system based upon the blackboard architecture. The blackboard architecture extends the common expert system design to include multiple expert systems, termed Knowledge Sources (KS's), which combine to solve problems too diverse or complex for conventional expert systems. Extending the architecture, MIMIR uniquely adds Data Sources (DS's) to the conventional KS's for problem decomposition and solution. Developed in Common LISP and CLOS, MIMIR can answer basic questions about the data in the remote databases and generate multiple queries for more complex questions. Seven partitions in MIMIR's blackboard allow KS's and DS's to focus on specific levels of the problem decomposition. MIMIR relies on an intelligent interface to translate an internal LISP-based SQL-like query into a valid SQL query string. This query is then be submitted to an external relational database and results are returned to the blackboard environment. While MIMIR is currently limited to SQL-accessible relational databases, the architecture can be extended to support interfaces to other data formats.

Pages

168

Share

COinS