针对产品设计时间预测中存在的小样本、不确定性数据等问题,将不确定数据处理成区间数,引入Hausdorff距离以计算区间数向量之间的距离,并将含有区间数的知识设计转换成核近似可嵌入的线性不等式,提出相应的产品设计时间智能预测方法,给出相关参数的优选方法。进行了注塑模具设计的实例分析,结果表明基于嵌入知识的核近似的时间预测方法是有效的。
Aiming at the problems of small samples and data uncertainty in product design time forecast,data uncertainty were transformed into intervals.Hausdorff distance was introduced to compute the distance between two interval vectors,and prior knowledge with intervals was incorporated into kernel approximation in the form of linear inequalities.An intelligent time forecast method and its relevant parameter-selection algorithm were put forward.Results of application in injection mold designs revealed that the time forecast method based on knowledge-incorporated kernel approximation was effective.