Application Research Of Artificial Intelligent Technology In Product Familiy-Based Product Definition

In the current scenario of globalization and strong competitive environment, mass customization (MC) has become an effective method for companies. It can satisfy wide spectrum of customer demands with low cost, which results in high profit. During the design stage of MC, product definition is one of the essential premises for successful product development. It requires design engineers to catch up with the voice of customers and translate it into product specifications effectively. The main concern of this paper is focused upon the following aspects:(1) Review the background of mass customization, including economic effect, research status and trend. Based on the discussion of the key issues in MC, the research approach is presented.(2) Review the research activities of product definition, including acquisition, representation and preprocessing of customer needs, representation and analysis of functional requirements, methodology of mapping individual customer requirement into product specifications, etc.(3) Propose a model of product family identification based on two kinds of knowledge. Using knowledge-based artificial neural network (KBANN), this model can learn from two knowledge resources and realize the mapping relationship between customer needs and product family. KBANN, combining the merits of inductive and deductive methods, can learn knowledge from data base and rule base. On the basis of KBANN, this paper utilize classification and regression tree (CART) to translate the knowledge store in KBANN into comprehensible rules, which can help engineers understand customer preference to different product families.(4) Propose a fuzzy intelligent design retrieving system for product definition. It employs fuzzy ARTMAP (FAM) neural network as its key technique to retrieve reference designs based on the similarity of customer requirements and functional requirements. FAM learns from historical transaction records. For newly emerging customer requirements, the system is able to effectively retrieve reference designs. Design engineers can then develop new products by evaluation and modification of the existing ones.(5) Present the framework of product specification identification base on data mining method. It mines the complex relationship between customer need attributes and functional attributes of product. Four objectives, support, confidence, interestingness and comprehensibility, are used for evaluating the extracted rules. To solve such a multi-objective problem, a Pareto-based GA is utilized to perform the rule extraction.Take customer's demand as the starting point of management, this article investigated the product definition in mass customization. It improves the design methodology of MC, reduces time and cost during the design process, and thus increases the profits of companies.