Constraint satisfaction problem (CSP) paradigm has proven highly successful in product configuration, particularly for build-to-order products, by assigning component types to all components without violating any constraints. For engineer-to-order products, however, product configuration requires assigning design parameters to each component as well. Hence, it often involves numeric variables, n-ary constraints, and constraints over variables that depend on other variables. Thus, an efficient search strategy is needed to address these issues. In this paper, an extension to the CSP, called Dependent CSP, is proposed to accommodate the complex engineer-to-order product configuration and the search strategy. In the Dependent CSP, variables are categorized as independent variables and dependent variables so that, search space can be reduced by eliminating dependent variables. Backjumping search strategy is employed to search for a solution as effective as possible. An updating mechanism is designed to avoid repetitive and unnecessary variable updating and constraint evaluation. Several variable ordering heuristics are assessed and the most effective ones are chosen for solution implementation. By applying these strategies, we can achieve a very efficient search algorithm for product configuration. The algorithm has been applied in a product configuration problem - an elevator system design - and a configuration solution can be obtained in a matter of seconds.
[1]
Markus Stumptner,et al.
An overview of knowledgedbased configuration
,
1997
.
[2]
Gregg R. Yost,et al.
Configuring elevator systems
,
1996,
Int. J. Hum. Comput. Stud..
[3]
Markus Stumptner,et al.
Configuring Large Systems Using Generative Constraint Satisfaction
,
1998,
IEEE Intell. Syst..
[4]
Roman Barták,et al.
Constraint Programming: In Pursuit of the Holy Grail
,
1999
.
[5]
Qiang Shen,et al.
Solution Techniques for Constraint Satisfaction Problems: Foundations
,
2001,
Artificial Intelligence Review.
[6]
Sanjay Mittal,et al.
Towards a Generic Model of Configuraton Tasks
,
1989,
IJCAI.
[7]
Rina Dechter,et al.
Backjump-based backtracking for constraint satisfaction problems
,
2002,
Artif. Intell..
[8]
Helen Xie,et al.
Towards Engineer-to-order Product Configuration
,
2002,
CAINE.
[9]
Brian Falkenhainer,et al.
Dynamic Constraint Satisfaction Problems
,
1990,
AAAI.
[10]
Boi Faltings,et al.
Solving Mixed and Conditional Constraint Satisfaction Problems
,
2003,
Constraints.
[11]
Markus Stumptner,et al.
A Generative Constraint Formalism for Configuration Problems
,
1993,
AI*IA.
[12]
Vipin Kumar,et al.
Algorithms for Constraint-Satisfaction Problems: A Survey
,
1992,
AI Mag..