目的 为解决光学元件表面进行离子束抛光加工过程中存在边缘效应的问题。方法 采用将驻留时间网格在面形采样网格的基础上进行延拓的方法来抑制边缘效应的产生，即将驻留点的网格范围向外进行延拓，使元件边缘部分的采样点中有去除作用的驻留点的数量与中间部分相同，延拓距离大于离子束半径。使用截断奇异值方法求解线性方程组模型的驻留时间，分析仿真结果中边缘效应出现的原因。结果 当驻留时间采用和面形误差同样的网格划分时，残留面形不收敛，且发生严重的边缘效应。当取截断参数k=80时，其面形PV值由初始的104.489 nm下降到11.675 nm，RMS值由28.009 nm收敛到1.572 nm，且没有边缘效应的产生。采用上述模拟结果作为实验设定参数，在平面光学元件上进行离子束抛光实验研究，抛光前后，PV值由102 nm降为37 nm，RMS由23 nm 降为2 nm。结论 实验结果验证了模拟结果的有效性，使用驻留时间网格延拓的方法可以避免边缘效应的产生，在此基础上进行离子束抛光，光学元件面形PV及粗糙度达到了非常好的收敛效果。

The work aims to solve the edge effect problem during the ion beam figuring on the surface of the optical elements. The method of extending the dwell time mesh on the basis of the surface shape sampling mesh was used to suppress the generation of edge effects, i. e. the mesh range of the dwell point was extended, so that the number of dwell points that removed the sample points at the edge portion of the component was the same as in the middle portion and the extension distance was greater than the ion beam radius. The truncated singular value method was used to solve the dwell time of the linear equations model, and the causes of the edge effects in the simulation results were analyzed. When the dwell time adopted the same meshing as the surface error, the residual surface did not converge and a serious edge effect occurred. When the truncation parameter k=80, its PV value decreased from the initial 104.489 nm to 11.675 nm, and the RMS value converged from 28.009 nm to 1.572 nm with no edge effect. With the above-mentioned simulation results as experimental setting parameters, ion beam figuring experiments were carried out on planar optical elements. Before and after figuring, the PV value decreased from 102 nm to 37 nm and RMS decreased from 23 nm to 2 nm. The experimental results basically verify the validity of the simulation results. The method of dwell time mesh extension can avoid the edge effect. On this basis, in the ion beam figuring, the PV value of the surface shape and the roughness of the optical components can achieve a very good convergence effect.