A knowledge-based approach to automatic generation of assembly plans
Abstract
In this research, the development of a knowledge-based assembly planning system is investigated. The rationale behind the development of such a system are two folds: one is that manual assembly planning is slow and should be automated in order to improve the efficiency of manufacturing system; the other is that automatic assembly planning can provide assemblability feedback to the designers so that design improvements can be made. A relational model has been proposed to describe assemblies in computers. Based on this relational model, assembly plans, both monotone and nonmonotone, can be described as an ordered list of relations to be established. The knowledge about the ordering constraints among relations is called precedence constraint knowledge. Two types of basic precedence constraints are identified, one is called must precede, and the other is called no later than. Any precedence constraint of an assembly can be described using these two basic relationships and their logical expressions. Using geometric reasoning techniques, algorithms have been developed to obtain the precedence constraint knowledge automatically from the CAD model of the assembly. In addition to knowledge about assembly itself, the effects of resource constraints are also modeled in terms of precedence constraints between the application of fixtures (or tools) and assembly operations. These constraints can be obtained through geometric reasoning techniques as well. The knowledge about assembly and assembly cell can be represented using predicate calculus, and forms the static knowledge database of the proposed knowledge-based assembly planning system. Rules have been developed to model the effects of disassemblying, change of fixture and change of tools. By applying these rules successively, all feasible assembly plans can be generated. The proposed assembly planning system has been implemented on a Silicon Graphics personal IRIS workstation. It provides both user-friendly interface and graphical simulation facilities to verify the generated assembly plans.
Degree
Ph.D.
Advisors
Lee, Purdue University.
Subject Area
Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.