According to the delay feature, the efficient driving for deformable mirror can simplify the structure of system. A deformable mirror control algorithm on the basis of Fully Convolutional Networks (FCN) is proposed. The control system including FCN can calculate the driving voltages of the deformable mirror according to the input signal, so as to realize the driving control of the deformable mirror. The results suggest that the algorithm simplifies the control system, and its calculation accuracy is better than the traditional Convolutional Neural Networks. At the same time, the present algorithm is suitable for removing the wavefront aberration.

中文翻译：

---

基于全卷积网络的变形镜简化控制方法

根据延迟特性，高效驱动可变形镜可以简化系统结构。提出了一种基于全卷积网络的可变形镜像控制算法。包括FCN的控制系统可以根据输入信号来计算可变形镜的驱动电压，从而实现对可变形镜的驱动控制。结果表明，该算法简化了控制系统，其计算精度优于传统的卷积神经网络。同时，本算法适用于消除波前像差。