Li Yongxiao, Wang Zinan, Peng Chao, Li Zhengbin
Appl Opt. 2014 Oct 10;53(29):6853-60. doi: 10.1364/AO.53.006853.
Conventional signal processing methods for improving the random walk coefficient and the bias stability of interferometric fiber-optic gyroscopes are usually implemented in one-dimension sequence. In this paper, as a comparison, we allocated synchronous adaptive filters with the calculations of correlations of multidimensional signals in the perspective of the signal subspace. First, two synchronous independent channels are obtained through quadrature demodulation. Next, synchronous adaptive filters were carried out in order to project the original channels to the high related error channels and the approximation channels. The error channel signals were then processed by principal component analysis for suppressing coherent noises. Finally, an optimal state estimation of these error channels and approximation channels based on the Kalman gain coefficient was operated. Experimental results show that this signal processing method improved the raw measurements' variance from 0.0630 [(°/h)2] to 0.0103 [(°/h)2].
用于提高干涉式光纤陀螺仪随机游走系数和偏置稳定性的传统信号处理方法通常在一维序列中实现。作为对比，本文从信号子空间的角度出发，通过多维信号相关性计算来配置同步自适应滤波器。首先，通过正交解调获得两个同步独立通道。接着，进行同步自适应滤波器操作，以便将原始通道投影到高度相关的误差通道和近似通道。然后，对误差通道信号进行主成分分析以抑制相干噪声。最后，基于卡尔曼增益系数对这些误差通道和近似通道进行最优状态估计。实验结果表明，这种信号处理方法将原始测量值的方差从0.0630[(°/h)²]提高到了0.0103[(°/h)²]。
