为快速、精确的进行空间直线度误差评定,提出了一种最小二乘算法与人工鱼群算法相结合的混合优化算法解决该问题。首先利用改进的最小二乘算法获取过测点集合算术平均中心的最小二乘拟合直线,然后在该直线向量邻域内均匀生成人工鱼群算法的初始解,进而基于旋转逼近策略应用改进后的人工鱼群算法搜索最小包容圆柱的轴线参数。通过在经典人工鱼群算法中引入变异和淘汰机制,对传统鱼群算法中的聚群、觅食等行为加以改进,有效提高了鱼群算法的优化效率和稳定性。实验及仿真结果表明:文中算法与遗传算法、粒子群算法等其它多种算法相比具有更高的正确度,非常适合空间直线度误差的精确评定。
In order to accurately and efficiently evaluate the straightness error, a hybrid optimization algorithm, combining the least squares algorithm and artificial fish swarm algorithm (LS-AFSA) was presented. First, the least squares line through the center of the measured points was obtained through the improved least squares algorithm and the initial solutions of AFSA were randomly generated in the neighboring line vector. Then the axis approaching the minimum enclosure cylinder was obtained by using the AFSA algorithm via rotation approaching strategy. A mutation and eliminated mechanism was applied in classic AFSA algorithm to improve the behavior of artificial fish and the precision and convergence speed of the AFSA were increased. Finally, the experiment al results show that the present hybrid optimization algorithm has higher accuracy than that via genetic and other algorithm, and does more appropriate for precisely evaluating the spatial straightness error.