摘要
采用人工智能算法依照最小区域法评定球度时,终止条件存在过度迭代和迭代不足问题。依据不确定度传递公式和误差可以忽略的条件推导得:球度评定时球心坐标的不确定度小于0.47倍测量不确定度时可以忽略。逼近程度小于这个值后,继续逼近是没有意义的。通过对网格算法评定球面度中网格容差的确定验证了该结论有效性。结论作为智能评定球度时的终止条件,比传统的以迭代次数和迭代时间、适应度限作为结束条件更为科学、合理。
There is always the average and insufficiency iterative in the arithmetic on the minimal area method algorithms of sphericity error. The derivation is made by uncertainty propagation and by errors which can be ignored. The center uncertainty can be ignored when the center uncertainty is less than 0.47 times the measurement uncertainty. It is worthless of continuing approximate when the assessment value is 0.47 times less than the measurement uncertainty. It is feasible by mesh tolerance determination in sphericity error based on mesh search algorithm it is more scientific and reasonable than termination of time or genetic algebra or fitness in the sphericity error assessment by genetic algorithm.
出处
《机械设计与制造》
北大核心
2014年第11期208-210,共3页
Machinery Design & Manufacture
基金
河南省教育厅青年骨干教师资助项目(2010GGJS-294)
关键词
网格容差
不确定度评定
球度误差
最小区域
网格算法
结束条件
Mesh Tolerance
Uncertainty Assessment
Sphericity Error
The Minimal Area Method
Mesh Search Al-gorithm
Termination