KNOWLEDGE-BASED FACILITIES DESIGN (EXPERT SYSTEMS, ARTIFICIAL INTELLIGENCE, MANUFACTURING)
Abstract
This dissertation presents a knowledge-based, expert system approach to the problem of facilities design. Facilities design as described in this work involves the determination of how components of a factory can best support achieving the objectives of the factory. This includes the selection, specification and layout of these manufacturing components. A prototype system, FADES, is presented to demonstrate the proposed approach. The development of FADES has resulted in the establishment of design models, specifications, and expert designer rules-of-thumb in a "knowledge-base" which can act upon user-specific design problems in achieving a "best" solution. FADES is programmed in PROLOG and runs under the UNIX operating system on a VAX 11/780. It is the intention of FADES to demonstrate expert-level capabilities in three main areas of facilities design: (1) Design problem definition and objectives. (2) Facilities selection and specification. (3) Facilities layout. Additionally, capabilities have been built into FADES which foster the user interface. These include a rule base editor, explanation of reasoning used, and recognition of user experience level. The FADES system has been tested using realistic design problems from cooperating companies with selected test results given. The test results indicate that the FADES approach can significantly enhance current facilities design capabilities in quality of solution, user communication, transparency of reasoning, and dealing with ill-structured problems. The research described also provides insight into issues of interest for other manufacturing problems, as regards the development of intelligent decision-support systems: (1) acquistion of manufacturing problem-solving knowledge. (2) manufacturing knowledge representation. (3) automatic data acquisition via the query language of a commercial data base management system, MDBS III. (4) combining judgmental, procedural, and declarative design knowledge with analytical models in the design process. (5) interfacing the system with existing models and algorithms written in different programming languages. (6) combining multiple expert functions in solving a problem, e.g., diagnosis, planning, and design. (7) combining multiple inference control strategies to obtain a problem solution, e.g., data-driven and goal-driven strategies for selected problem components.
Degree
Ph.D.
Subject Area
Industrial engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.