采用遗传算法提高自由曲面光顺特性

为了提高自由曲面的光顺性，对采用遗传算法进行曲面光顺的算法机理、实现过程和平滑特征展开研究。首先利用模糊罚函数拟合激光扫描获取的离散点，然后根据群体规模在变量取值空间随机产生初始群体，计算群体中各个个体的标准目标函数值和约束函数值，最后根据模糊罚函数法将原来带约束的多目标优化问题转化为无约束多目标优化问题，并根据群体中的点的适应度值进行选择。实验结果表明：采用遗传算法进行曲面光顺处理，相比三次B样条拟合，曲面反射线分布更加均匀，自由曲面最大曲率变化率减小0.1。

Improvement of surface fairing with genetic algorithm

In order to improve the fairing property of free surface,the algorithm principle,realizing procedure and fairing property of the genetic algorithm are studied.First,ambiguity punishment function is adopted to fit point clouds obtained from laser scanning.Then,original group are produced at random in variable space corresponding to group scale,and the standard target function value and constraint function value of each individual among the group are calculated.Finally,multi-target optimization with constraint is transformed to multi-target optimization without constraint according to the method of ambiguity punishment function,and the next generation is chosen with respect to the adaption degree value of the point.Experimental results indicate that compared to 3-order B-spline fitting,the reflection lines of the free surface by genetic algorithm faring are smoother,and the maximum curvature variance ratio decreases by 0.1.