%0 Journal Article
%T 基于深度学习的分数阶涡旋光拓扑荷和传输距离的双重识别<br>Dual Identification of Fractional Optical Vortices Topological Charge and Transmission Distance Based on Deep Learning
%A 谢超
%A 吴凡
%A 戈峰
%A 刘清越
%A 王冠
%J Modeling and Simulation
%P 88-96
%@ 2324-870X
%D 2025
%I Hans Publishing
%R 10.12677/mos.2025.142134
%X 分数阶涡旋光与整数阶涡旋光相比具有更加灵活和复杂的轨道角动量等物理特性&#65292;在多领域有着非常广泛的应用。近年来&#65292;有相关报道在实验中有效同时识别了整数阶涡旋光的拓扑荷和传输距离&#65292;为涡旋光的识别提供了新的方向。相比之下&#65292;分数阶涡旋光的双重识别研究相对较少&#65292;并且随着湍流的加强&#65292;相邻整数阶涡旋光在识别中易发生混叠。为针对这一问题&#65292;本文在基于大气湍流和高斯白噪声的条件下&#65292;提出了一种使用深度学习的方法来识别分数阶涡旋光的拓扑荷和传输距离的方案。本文使用残差卷积神经网络&#65292;并使用改进的非Kolmogorov大气湍流模型&#65292;在仿真对比中&#65292;不同强度大气湍流和高斯白噪声的条件下对距离间隔为100米的分数阶涡旋光光强分布图像进行识别&#65292;识别准确率比整数阶涡旋光高5%以上&#65292;在距离间隔为100米、50米的不同大气湍流强度和高斯白噪声条件下都得到了比较高的识别精度。<br>Fractional optical vortices, compared to integral optical vortices, possess more flexible and complex physical properties, such as orbital angular momentum, leading to widespread applications across various fields. In recent years, experimental studies have successfully achieved the simultaneous identification of the topological charge and transmission distance of integral optical vortices, providing new directions for optical vortex recognition. In contrast, research on the dual identification of fractional optical vortices remains relatively limited. Moreover, as turbulence intensifies, adjacent integral optical vortices are prone to entanglement during recognition. To address this issue, this paper proposes a deep learning-based method for identifying the topological charge and transmission distance of fractional optical vortices under conditions of atmospheric turbulence and Gaussian white noise. A residual convolutional neural network was employed, along with an improved non-Kolmogorov atmospheric turbulence model. In simulation comparisons, the intensity distribution images of fractional optical vortices at a transmission distance of 100 meters were identified under varying strengths of atmospheric turbulence and Gaussian white noise, achieving an identification accuracy over 5% higher than that of integral optical vortices. Furthermore, high recognition accuracy was achieved under different atmospheric turbulence strengths and Gaussian white noise conditions at transmission distances of 100 meters and 50 meters.
%K 分数阶涡旋光&#65292
%K 深度学习&#65292
%K 大气湍流&#65292
%K 拓扑荷&#65292
%K 传输距离<br>Fractional Optical Vortices
%K Deep Learning
%K Atmospheric Turbulence
%K Topological Charge
%K Transmission Distance
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=107782