关键词: 产品族, 配置性能预测, 支持向量机, 混合蛙跳算法

Abstract: To predict the configuration performances of modular product family design accurately and quickly for judging whether to meet the diverse customer demand,a configuration performance prediction method of modular product family based on Hierarchical Support Vector Machine (H-SVM) was proposed.The new variant product performances were predicted through the data mining on the history database of typical product in the product family.The framework and basic procedure of configuration comprehensive performance prediction method based on H-SVM was introduced.An improved Shuffled Frog Leaping Algorithm (SFLA) was adopted to optimize the kernel function parameters and error penalty factors of SVM models,where the simulated annealing algorithm was used to increase local searching ability and convergence velocity.The method was verified on a developed modular paper currency sorter product family to assess the effectiveness of the proposed method.

Key words: product family, configuration performance prediction, support vector machine, shuffled frog leaping algorithm