论文部分内容阅读
分振幅光偏振仪是一种测量速度快的光波传感器,而数据处理是分振幅光偏振仪的应用基础,针对标准神经网络存在的缺陷,提出一种改进神经网络的分振幅光偏振仪数据处理方法。采用分振幅光偏振仪的电信号作为输入,入射光斯托克斯参数作为期望输出,采用神经网络拟合输入与输出之间的关系,K聚聚类算法对神经网络参数进行优化,对分振幅光偏振仪数据处理的测试实验结果表明,本文方法获得了较高精度的入射光斯托克斯参数估计结果,性能要优于传统方法。
The polarization amplitude polarimeter is a kind of optical wave sensor with high measuring speed. The data processing is the basis of polarization amplitude polarimeter. Aimed at the defects of the standard neural network, this paper proposes an improved neural network polarization amplitude polarimeter data processing method. The electrical signal of the polarization amplitude polarizer is used as input and the incident light Stokes parameters as the expected output. The neural network is used to fit the relationship between input and output. K-clustering algorithm is used to optimize the neural network parameters. Experimental results of amplitude light polarizer data processing show that the proposed method achieves better results of Stokes parameter estimation of incident light, and its performance is better than the traditional method.