Guided genetic algorithms for solving a larger constraint assembly problem

Abstract

Assembly planning calls for the subtle consideration of certain limitation factors such as geometric features and tools so as to work out a specific assembly sequence. From the assembly sequence, all parts will be turned into a product. It is evident that the degree of complexity of the assembly problem will increase when the number of constraints is larger. Using Genetic Algorithms (GAs) to solve the assembly sequence features speed and flexibility can fit the requirements of various domains. In the case of larger constraint assembly problems, however, GAs will generate a large number of infeasible solutions in the evolution procedure, thus reducing the efficiency of the solution-searching process. Traditionally, using GAs is a random and blind-searching procedure in which it is not always the case that the offspring obtained through the evolutionary mechanism will meet the requirements of all limitations. In this study, therefore, Guided-GAs are proposed wherein the proper initial population and the alternation of crossover and mutation mechanisms are covered to overcome assembly planning problems that contain large constraints. The optimal assembly sequence is obtained through the combination of Guided-GAs and the Connector-based assembly planning context as previously suggested. Finally, practical examples are offered to illustrate the feasibility of Guided-GAs. It is found that Guided-GAs can effectively solve the assembly planning problem of larger constraints.

Keywords:

• Connectors
• Guided genetic algorithm
• Binary tree
• Guided crossover
• Guided mutation

Acknowledgements

Research was supported by the National Science Council of the Republic of China under Grant Number NSC 93-2213-E-167-023.