| Today, product is updated rapidly, and the enterprises have to innovate to face intense market competition. A product evolution method can help enterprises to grasp market trends and improve the speed of product innovation. Extension case reasoning is a product evolution method which is used to solve design problem by previous design knowledge, it is a design method based on case. Case representation, case matching, case modification and case evaluation are main components of case reasoning. In this paper, case classification is added as a component, and case retrieval is based on dependent multidimensional function, the main research contents are as follows:Case representation and case matching are the base of product design evolution, in this paper, the information model of product cases and customers’ requirement is established based on basic-element, and extension field and the optimal value is computed according customers’ requirement. Then a matching algorithm of case matching for product design evolution is proposed, mathematical model of dependent multidimensional function is established through summarized algorithm of dependent two-dimensional function and dependent three-dimensional function. On the base, the similarity of product case is computed. At last, the new algorithm is verified it’s reliable and correctness through case application and comparison.To improve efficiency of product design evolution and make evolution object clearer, a product classification based on back propagation(BP) network is proposed. According to characteristics of product classification, error calculation method is optimized to improve convergence rate of BP network. And training samples are pretreated to improve accuracy of product classification. Finally, according to similarity, the number of mismatching attributes and mismatching sub-attributes, product cases are classified through this BP network.Product design evolution realization is depended by case modification. Firstly, mapping model between product attribute and structure is established, and the incompatible core problem of similar case is got by this model. Secondly, According to substitution transformation, duplication transformation, composition/decomposition transformation, increasing/decreasing transformation, expansion/contraction transformation and propagation degree, the structure or dimension of product can be modified to meet customers’ requirement.A screw air compressor design system based on extension case reasoning is developed on Solidworks according to the theory by technology and method which is put forward. Retrieval algorithm based on dependent multidimensional function, back propagation network classification and incompatible problem solving method are integrated in this system, and the system proved the theory is feasible and viable. |
PDF Full Text Request