Fund Project:Project supported by the National Natural Science Foundation of China (Grant Nos. 51675380, 51775379), the National Key R&D Program of China (Grant No. 2018YFB2003501) and the Key Technologies R&D Program of Tianjin, China (Grant No. 18YFZCGX00920)
Received Date:29 May 2019
Accepted Date:07 July 2019
Available Online:01 October 2019
Published Online:05 October 2019
Abstract:With the rapid development of modern technology, high-precision absolute distance measurement is playing an important role in many applications, such as scientific research, aviation and industry measurement. Among the above various measurement methods, how to realize higher-accuracy, larger-scale, and faster-speed measurement is particularly important. In the traditional technique for long-distance measurement, the emergence of optical frequency comb (OFC) provides a breakthrough technology for accurately measuring the absolute value of distance. The OFC can be considered as a multi-wavelength source,whose phase and repetition rate are locked. The OFC is a very useful light source that can provide phase-coherent link between microwave and optical domain, which has been used as a source in various distance measurement schemes that can reach an extraordinary measurement precision and accuracy. A variety of laser ranging methods such as dual-comb interferometry and dispersive interferometer based on femtosecond laser have been applied to the measuring of absolute distance.In this paper, the factors affecting the resolution and the non-ambiguous range of spectral interferometry ranging using OFC are particularly discussed. We also analyze the systematic errors and the limitations of traditional transform methods based on Fourier transform, which can conduce to the subsequent research.To address the problem caused by low resolution and unequal frequency interval, we propose a data processing method referred to as equal frequency interval resampling. The proposed method is based on cubic spline interpolation and can solve the error caused by the frequency spectrum broadening with the increase of distance. Moreover, we propose a new method based on least square fitting to calibrate the error introduced by the low resolution of interferometry spectrum obtained with fast Fourier transform (FFT). With the proposed method, the simulation results show that the systematic error is less than 0.2 μm in the non-ambiguity range and the system resolution is greatly improved. Finally, anabsolute distance measurement system based on Michelson interferometer is built to verify theproposed method. The measurement results compared with those obtained by using a high-precision commercial He-Ne laser interferometer show that the distance measurement accuracy is lower than 3 μm at any distancewithin the non-ambiguity range. The experimental results demonstrate that our data processing algorithm is able to increase the accuracy of dispersive interferometry ranging with OFC. Keywords:optical frequency comb/ frequency domain interferometer/ spectral interferometry/ absolute distance measurement
全文HTML
--> --> -->
2.光频梳频域干涉测距原理光学频率梳在频域上可以提供一系列分立的偏移频率为fCEO, 频率间隔为frep的频率纵模[11], 如图1所示. 图 1 光频梳频谱 Figure1. Spectrum of optical frequency comb
时域上表现为一系列等间隔的飞秒级脉冲序列, 时间间隔${T_{\rm{R}}} = 1/{f_{{\rm{rep}}}}$, 群速度和相速度差异产生的载波包络相位偏移$\Delta {\varphi _{{\rm{ce}}}} = {{2{\text{π}}{f_{{\rm{CEO}}}}} / {{f_{{\rm{rep}}}}}}$. 由于光频梳具有相位锁定特性, 相邻脉冲间的相位关系固定, 使不同脉冲之间可以相互干涉[16]. 频域干涉法又叫光谱干涉法或色散干涉法, 测量系统结构为经典Michelson干涉仪[11,19], 如图2(a)所示. 其中参考镜Ref固定, 测量镜Target可前后移动以改变光程差2L, 光频梳发出的脉冲经分束镜BS分束后被分别反射并产生时间差τ, 再次汇聚于BS时在频域内发生干涉, 光谱仪接收干涉信号并输出结果. 图 2 (a)频域干涉法原理图; (b) 测距系统实物图 Figure2. (a) Principle of frequency domain interference method; (b) the experimental setup of the measurement system
${L_{{\rm{NAR}}}}$影响因素已由前文推导, W = 1510—1610 nm, $\Delta \lambda = 0.01\; {\rm{ nm}}$, 通过分析波形展宽并将测量距离与干涉仪结果相对比, 验证等频率间隔重采样和峰值位置拟合法的误差修正和对测量范围的提升效果. 由图10(a)—(c)可知传统FFT法的误差累积明显, 超过${{{L_{{\rm{NAR}}}}} / 2}$时波形展宽已不能正常计算, 本算法有效修正了该问题, L较大时仍保持明显的窄脉冲. 与干涉仪结果相比对以验证修正效果, 见表2. 图 10 峰值定位效果对比仅传统FFT (a) L = 5.8600 mm, (b) L = 16.9850 mm, (c) L = 27.9100 mm; 等频率间隔重采样和峰值位置拟合(d) L = 5.8600 mm, (e) L = 16.9850 mm, (f) L = 27.9100 mm Figure10. Effect contrast of peak position fitting. Original FFT: (a) L = 5.8600 mm, (b) L = 16.9850 mm, (c) L = 27.9100 mm. Equal frequency interval resampling and peak position fitting: (d) L = 5.8600 mm, (e) L = 16.9850 mm, (f) L = 27.9100 mm
实验序号
L/mm
传统FFT法误差/μm
等频率间隔重采样误差/μm
峰值拟合误差/μm
1
0.0023
–2.3015
1.6570
0.0218
2
0.0037
–3.6001
3.0337
–0.7335
3
0.1000
3.2167
0.1031
0.8437
4
0.4997
–1.8309
0.4975
0.2267
5
0.9998
2.0097
1.0011
–0.9828
6
3.1307
–4.2793
3.1336
1.1003
7
4.9980
–13.2375
4.9995
–0.1739
8
6.2364
157.3300
6.2308
–1.9198
9
6.2511
无法定位
6.2551
–2.9791
10
9.3629
无法定位
1.5705
–2.1087
表2传统FFT、等频率间隔重采样和峰值位置拟合法误差比较 Table2.Measurement results of different distance.