圆度误差评定与测量不确定度计算

Evaluation of Circularity Errors and Computation of Measurement Uncertainty

为了更准确地评定圆度误差及测量不确定度,根据圆度特点,提出实数编码改进遗传算法求圆度误差最小区域解,基于蒙特卡洛法评定测量不确定度.通过对零件实测计算,结果表明采用实数编码的改进遗传算法不仅省去了重复的编码解码,而且算法简单、优化效率高,蒙特卡洛法计算不确定度与传统GUM方法相比不受直接测量量相关性的限制,而且受问题条件限制的影响小,使不确定度评定简单化.采用改进遗传算法和蒙特卡洛法能够更加准确高效地评定圆度误差和测量不确定度.

For a more accurate evaluation of circularity errors and measurement uncertainty, this paper suggests an improved real-code genetic algorithm using Monte Carlo method. The results of calculation of actual parts indicate that the modified algorithm is simple to apply and saves the efforts for repeated encoding and decoding, with high efficiency in optimization. Compared with traditional GUM method, Monte Carlo method is not restricted by direct measurement quantity correlation, and less likely to be conditioned by problem constraint, thus simplifying uncertainty evaluation. Consequently, the improved genetic algorithm and Monte Carlo method can be applied to accurately evaluate circularity error and measurement uncertainty with high efficiency.