Research On Constraint Processing And Solving Method Of Product Configuration Problems

Compared to the past, the manufacture has some new trends today: production of mass customization (MC), the globalization of trade, the individuation of customer demands. Under the various pressure of competition, enterprises have to develop products that using newer technologies and have more functional. In addition, customers want more and individual demands of the products, therefore, enterprises must increase the complexity of the product, shorten the product life cycle, minimize product "sales-delivery" time, and to provide customers with personalized products.Configurable product is a new concept that can be used to meet these new trends. The configurable product includes some pre-defined general parts and the constraints of these parts, so on the premise of meeting the constraints we can assemble different parts and satisfy customers demands. By designing configurable products and configuring configurable products on sales process, enterprises can get produce products which satisfy customer individual demands.Therefore, the product configuration is a process of assembling pre-defined parts of configurable product which in accordance with the constraints and meeting the user's requirements. Product configuration system—Product Configurator, which constructed by computer, can automatically and rapidly complete complex product configuration tasks, the efficiency and correct of configuration can not achieve by hand. From the perspective of artificial intelligence, product configurator is equivalent to an expert system: product configurator stores configuration knowledge in the knowledge base, and uses certain methods reasoning the rules according to customers'demands and returning the configuration results to users.Constraint-based product configuration regards the parameters of general product as variables which have different values. Different parameter values for different types of parts, the values of parameters are the result ofconfiguration which meet all the constraints. Therefore, Constraint-based product configuration will be converted to product configuration knowledge, and the solving process will be converted to the search until the values of parameters satisfy all the constraints. In this paper, we use constraint satisfaction problem (CSP) to express Constraint-based product configuration and solve the problem. Due to constraints and variables can be used to describe the configuration knowledge conveniently, meanwhile constraint satisfaction problem with a variety of effective solution, therefore it is easier to describe the configuration problems of high efficiency.Product configuration problems constraints are put into two categories: general product constraints and user constraints. General product constraints are build by modeling expert in building product model, these are restrictions of general parts or restrictions of parameters; user constraints are build in the configuration of individual products, which set up the parameter value required. Many of the configuration systems are not very good at input the constraints, one key drawback is that experts should have deeper understanding of the constraints. Under normal circumstances, only experts in their fields of knowledge modeling a better understanding of the complex logic can not handle the constraints. Considering the characteristics of general products'constraints, in this paper, general products constraints fall into three categories: attribute parameter constraints, choosing parameter constraints, attribute and choosing parameter constraints, and given formal definition.Modeling experts is due to the importation of "IF condition THEN results" type of rule-based form of constraints. Attribute parameters directly to the preservation of such binding rules-based forms, will affect the maintenance of knowledge base in the future, so these constraints will be convert to other form when experts modeling general products. After it is not only easy to maintain and easy to solve based on the constraint satisfaction problems.Cardinality rules are bound to deal with such logic, "If produce A, B orC must also be produced". Proposed that cardinality rule can be converted into an equivalent linear inequalities group. Choosing parameter constraints can be transformed into cardinality rules. Because cardinality rules have the structure of a unified form, choosing parameter constraints are convenient input into knowledge base after convert into cardinality rules. After modeling experts input the constraints into system, cardinality rules are converted into inequalities and then the rules. This is to facilitate the configuration of products based on the constraint satisfaction problem solving.Product configuration problems which based constraint satisfaction problems are non-binary constraint satisfaction problems. Although the consistency technologies are used for non-binary constraint satisfaction problems, however, certain non-binary constraint satisfaction problems converted into binary CSP then solving is a good choice. Due to the uniqueness constraints and product configuration problems, transform them into binary CSP solving than binary CSP solving efficiency is much higher. Based on the practical application we select the efficient configuration algorithm. In this paper we give solving methods to assemble to order and design to order product configuration problems. The solving method for product configuration problem has been applied to product configurator Config2006, solution efficiency has been improved significantly than previous versions (Config2004). Assemble to order product configurator Config2006-II currently under development.Product configuration has broad prospects in the future. Product configurator will be available to users quickly and correctly for proper configuration and detailed explanation of the results, will be interact with users conveniently, will be simpler and easier to maintain product configuration knowledge of evolution, will be help enterprises and other products and information systems of enterprises seamless connectivity, will be ensure the correct application of the process of enterprise information systems. Enterprises, vendors and users will get on the bridge of communication.