A methodology for the automatic generation of process plans in an electronic assembly environment
Abstract
In this research important issues in the automatic generation of process plans are addressed. Since the conventional method of electronic assembly planning is subjective and depends greatly on the skills, memory, knowledge and experience of the planners, inconsistencies in the plan arise which in turn create problems in the manufacturing environment. An automatic electronic assembly planning system that would offer advantages over manual systems has been implemented. Due to frequent advances in technology the electronic industry is a very dynamic one. This affects the effectiveness of the current process plans and the applicability of the system that generates them. The framework of a system which takes into account the dynamic nature of the environment is presented as an example. A network representation of plans that allows for flexibility and modularity in the planning system is being used. A hybrid architecture consisting of production rules, procedural algorithms and a hierarchical database has being implemented using an expert system shell. The developed prototype system, called SAPIENT for System for Alternative Planning In Electronic-Assembly using a Network Technique, is a knowledge based system to automatically generate process plans for printed circuit boards (PCB) for small batch flexible line production. This is not a limitation of the system but rather a feature which allows the system to be more flexible. SAPIENT represents components, equipment capability and product design using a frame representation which allows property inheritance and is easy to maintain. The planning knowledge is represented by rules which allows expansion of new knowledge. The system has been tested with realistic data. The results of these tests proved that the suggested planning methodology works well in a flexible manufacturing environment.
Degree
Ph.D.
Advisors
Chang, Purdue University.
Subject Area
Industrial engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.