为了求解基于微机械薄膜变形镜自适应光学系统的最优模式控制电压，以使系统的入射畸变波前像差降到最小，采用了一种基于改进的奇异值分解闭环实时模式复原算法，可通过调整控制参量g₁，g_θ和W来优化波前复原的复原精度和收敛速度；采用一人眼出射畸变波前像差作为自适应光学系统的原始波前，与现有的几种实时模式复原算法相对比，其在收敛速度和复原精度上都有所改善。结果表明，基于改进的奇异值分解闭环实时模式复原算法，在一定程度上解决了微机械薄膜变形镜由于薄膜面型宽交连值、驱动电极单调但非线性的控制电压求解问题，算法可调整空间大、适用范围也较广。

An close-loop real-time algorithm of improvement singular value decomposition(SVD)modal reconstruction of wavefront was introduced in order to solve the optimal control voltages of adaptive optics based on the micromachined membrane deformable mirror(MMDM)and reduce the incident wavefront aberration to the minimum. By adjusting the control parameters such as g₁,g_θand W,reconstruction precision and convergence speed of wavefront were optimized. Wavefront aberration from the human eye was introduced as the incident wavefront of adaptive optics,and the reconstruction precision and convergence speed were improved by comparing the existent algorithms of wavefront reconstruction. As a result,the close-loop real-time algorithm of the improved SVD to some extent settled the control voltages of MMDM,which has the characteristic of great intercourse membrane and monotony but nonlinear driving electrode. This algorithm has good application and flexibility.