关键词: 参数标定, 产品变型主模型, 多目标优化, 容差划分, 田口式三步设计

Abstract: In order to improve the validity of variant product，calibration and tolerance identification to the parameters of master variant model is proposed. Aiming at calibration issue，quality characteristic indicators associated with customization product and its synthesis model are constructed, and then the range of key parameters attained from customized product instances is used as constraints. Based on these，parameter calibration value that makes synthesis indicator achieve its optimization is worked out by using multi-objective optimization genetic algorithm. Aiming at tolerance identification，the tolerance range of parameter is calculated within the allowance range of synthesis indicator by analyzing the product cost loss and indicator excursion on the basis of the three stage design of Taguchi method. As a result，a validity variant instance that satisfies the requirement of quality characteristics can be derived from the master product variant model，so the ratio of successfully deriving variant instance is increased. Finally，an example is given to verify the above-menotioned principle and method.

Key words: Master product variant model, Multi-objective optimization, Parameter calibration, Three stage design of Taguchi method, Tolerance identification