蒋立辉, 章典, 陈星, 熊兴隆. 基于谱估计的Fabry-Perot标准具透过率曲线参数估计方法[J]. 红外与激光工程, 2016, 45(9): 906006-0906006(8). DOI: 10.3788/IRLA201645.0906006
引用本文: 蒋立辉, 章典, 陈星, 熊兴隆. 基于谱估计的Fabry-Perot标准具透过率曲线参数估计方法[J]. 红外与激光工程, 2016, 45(9): 906006-0906006(8). DOI: 10.3788/IRLA201645.0906006
Jiang Lihui, Zhang Dian, Chen Xing, Xiong Xinglong. Parameter estimation method of Fabry-Perot etalon based on modern spectrum estimation[J]. Infrared and Laser Engineering, 2016, 45(9): 906006-0906006(8). DOI: 10.3788/IRLA201645.0906006
Citation: Jiang Lihui, Zhang Dian, Chen Xing, Xiong Xinglong. Parameter estimation method of Fabry-Perot etalon based on modern spectrum estimation[J]. Infrared and Laser Engineering, 2016, 45(9): 906006-0906006(8). DOI: 10.3788/IRLA201645.0906006

基于谱估计的Fabry-Perot标准具透过率曲线参数估计方法

Parameter estimation method of Fabry-Perot etalon based on modern spectrum estimation

  • 摘要: 直接探测多普勒测风激光雷达系统大多采用基于Fabry-Perot(F-P)标准具的条纹技术和边缘技术测量多普勒频移。但激光器频率漂移会使得透过率曲线的估计出现一定误差,为了抑制其影响,提出了一种基于谱估计的F-P标准具透过率曲线参数估计方法。首先将标准具全量程扫描结果进行重构得到观测矩阵。之后使用MUSIC算法得到透过率曲线的伪谱。对伪谱进行谱峰搜索后,将得到的谱峰位置进行线性拟合得到自由谱间距。然后将上步的拟合结果代入透过率函数,利用非线性最小二乘法对其他参数进行估计。在仿真分析的基础上,采用真实测量数据进行了实验验证。结果表明所提出的方法在扫描步点间隔达到50、信噪比大于10 dB的情况下,估计差小于1%。该算法扫描时间短,估计误差小,具有实际应用价值。

     

    Abstract: The Fabry-Perot(F-P) etalon was commonly used in direct detection Doppler wind lidar. As a considerable affect would be caused by laser frequency drift in transmittance curve estimation, a new transmission curve fitting algorithm based on spectrum estimation was presented. Firstly, the scan result of F-P etalon was reconstructed into observation matrix. The pseudo-spectrum by MISIC algorithm was conducted. Then spectrum peak was searched and the FSR by linear least squares estimation method was calculated. Secondly, the else parameters by nonlinear least square estimation method was estimated with the estimation above. The performance of the proposed method was analyzed by simulation and verified through the real data, the results show that the proposed algorithm is more stable and accurate under limited amount of data. The relative error is less than 1% when SNR is 10 dB and sample interval is 50 step. This method has lower sweep time, less error and realty utility value.

     

/

返回文章
返回