<p>The benefits of automating assembly sequence generation include: (1) insuring that no potentially good assembly sequence is overlooked, (2) reducing planning costs, (3) accelerating the analysis of the economical impact of different design solutions, (4) standardizing and improving the quality of the produced plans, and (5) contributing to achieving autonomous assembly systems. Previous research in assembly planning focussed on the generation and evaluation of all possible assembly plans for the product under consideration. This thesis presents a graph-theoretic approach for simultaneously generating and evaluating products' assembly/disassembly sequence alternatives and producing an optimum assembly plan according to predefined criteria. It aims at improving the efficiency of the assembly planning process and producing optimal assembly/disassembly plans. The developed graph-theoretic approach enables the determination of assembly sequences which transform any arbitrary initial state of the product into any arbitrary final state. Practically, this means many different types of assembly problems to be handled uniformly. A product is described in terms of its components and the assembly relationships between them. This description lends itself to a graph representation, where vertices correspond to the set C of assembly components and edges correspond to the set R of assembly relationships. For a product with "n" components, the generation of an assembly sequence is mapped into the problem of finding a sequence of "n-1" mutually exclusive cutsets in the graph model and its subgraphs. Each cutset corresponds to a disassembly operation of the physical product which produces two smaller subassemblies. Assembly sequences are encoded in a directed graph of assembly states, representing the search space. Geometric feasibility and accessibility constraints have been developed to help reduce this combinatorial search space. Assembly-related criteria which guide the search to an optimal solution are described. They are: (1) the number of re-orientations, (2) parallelism among assembly operations, (3) stability of subassemblies, and (4) clustering of similar assembly operations. Integrating the evaluation of these criteria as the search graph gets expanded, enables the direct generation of an optimal disassembly sequence of a given product with respect of these criteria. Standard search methods, including breadth first, depth first, best first, A* and hill climbing, are used to guide the search towards a single and optimal assembly sequence. The A* method can generate optimal solutions without explicitly generating the whole directed graph of assembly states. An interactive computer tool, based on the above approach, was developed. GAPP--a Generative Assembly Process Planner uses various search methods to incrementally construct the directed graph of assembly states and generate optimal assembly/disassembly sequences. Examples of real products are included to demonstrate GAPP's use and potential for assessing assembly, disassembly, repair, maintenance, assembly of multiple products and assembly error recovery procedures.</p> / Doctor of Philosophy (PhD)
| Identifer | oai:union.ndltd.org:mcmaster.ca/oai:macsphere.mcmaster.ca:11375/8573 |
| Date | January 1992 |
| Creators | Laperrière, Luc |
| Contributors | El, Hoda A., Mechanical Engineering |
| Source Sets | McMaster University |
| Detected Language | English |
| Type | thesis |
Page generated in 0.0016 seconds