Yue Dan, Song Pengcheng, Wang Chongshuai, Chuai Yahui
College of Physics, Changchun University of Science and Technology, Changchun 130022, China.
Sensors (Basel). 2023 Oct 12;23(20):8399. doi: 10.3390/s23208399.
To address the difficulty and complexity of detecting piston errors for segmented telescopes, this paper proposes a new piston error measurement method based on a hybrid artificial neural network. First, we use the Resnet network to learn the mapping relationship between the focal plane degradation image and signs of the piston error. Then, based on the established theoretical relationship between the modulation transfer function and the piston error, a BP neural network is used to learn the mapping relationship between the and the absolute value of the piston error. After the training of the hybrid network is completed, a wide-range and high-precision detection of the piston error of the sub-mirrors can be achieved using the combined output of the two networks, where only a focal plane image of the point source with broadband illumination is used as the input. The detection range can reach the entire coherent length of the input broadband light, and the detection accuracy can reach 10 nm. The method proposed in this paper has the advantages of high detection accuracy, a wide detection range, low hardware cost, a small network scale, and low training difficulty.
为了解决拼接望远镜活塞误差检测的困难和复杂性，本文提出了一种基于混合人工神经网络的新型活塞误差测量方法。首先，我们使用Resnet网络学习焦平面退化图像与活塞误差特征之间的映射关系。然后，基于调制传递函数与活塞误差之间建立的理论关系，使用BP神经网络学习调制传递函数与活塞误差绝对值之间的映射关系。在混合网络训练完成后，利用两个网络的联合输出，可以实现对子镜活塞误差的宽范围、高精度检测，其中仅使用宽带照明点源的焦平面图像作为输入。检测范围可以达到输入宽带光的整个相干长度，检测精度可以达到10纳米。本文提出的方法具有检测精度高、检测范围宽、硬件成本低、网络规模小和训练难度低等优点。
