Objective The distributed fiber-optic strain sensor based on the spectra of Rayleigh backscattering can detect static and dynamic signals. In addition, it is immune to interference fading. The Rayleigh scattering spectra of the optical fiber can be obtained using single-shot detection with frequency chirped pulses and matched filters; this solves the contradiction between spatial resolution and sensing distance. The frequency shift detection of Rayleigh scattering is the key in strain sensing. In this study, a Rayleigh frequency shift detection algorithm based on the frequency axis is proposed and verified in a configuration with chirped pulses and matched filters. Compared with the Rayleigh frequency shift detection algorithm on the distance axis, the signal-to-noise ratio is improved by more than 6 dB in dynamic strain measurement.
Methods The proposed system is based on a time-gated digital optical frequency domain reflectometer (TGD-OFDR) configuration (Fig. 1). Light wave from a narrow line-width fiber laser is split into local light and probe light. A filter is used to select the required sideband and suppress high-order harmonics. An erbium-doped fiber amplifier is used to boost the power of the probe pulse. The beat frequency between the Rayleigh backscattering and local oscillator from the balanced photo-detectors is sampled using an oscilloscope. Then, 91 matched filters with a frequency range and interval of 100 MHz and 5 MHz, respectively, are used to filter the beat signals, by which the Rayleigh scattering curves of different frequencies are obtained. At the position where the strain occurs, the spectral shift of Rayleigh backscattering on the frequency axis can be demodulated using a cross-correlation algorithm. The corresponding strain changes can be calculated using this frequency shift. Further, the strain waveform is obtained.
Results and Discussions The proposed distributed dynamic strain intensity-demodulation system successfully detects the strain on the optical fiber. By calculating the spectra of different pulses at the same position, the strain changes between pulses can be obtained (Fig. 2). To reduce the frequency interval, cubic spline interpolation is used before the cross-correlation algorithm for better strain resolution (Fig. 3). In the demonstrational experiment, 200 chirped pulses were launched into the fiber in sequence. The frequency-sweeping range of the probe pulse is from 200 MHz to 750 MHz. The width of each pulse is 4 μs, and the measurement period is 100 μs. The sinusoidal vibration with an amplitude of 100 nε applied to the far end of the 10 km fiber was successfully retrieved. The system’s spatial resolution is evaluated by the distribution of the standard deviation of the frequency shift. The distance between 10% and 90% of the rising edge of the vibration area is approximately 1 m, which is the system’s spatial resolution (Fig. 4). Compared with the intensity-demodulation method based on the distance axis, in this system, signal-to-noise ratio is improved from 25.7 dB to 32.5 dB and the frequency resolution is improved from 237 pε/Hz to 157 pε/Hz. The main reason for performance improvement is the usage of more Rayleigh feature information. Moreover, data processing time is reduced from 4.5 s to 0.8 s (Table 1).
Conclusions In this study, a distributed strain-sensing technique based on frequency shift of Rayleigh spectrum is proposed. The use of chirped pulses and matched filters eliminates the contradiction between spatial resolution and dynamic range, where the system’s spatial resolution is obtained using the frequency range instead of the pulse width. Compared with the spectral demodulation on the distance axis, the proposed data processing algorithm can reduce the amount of calculation and make full use of Rayleigh graphic information to improve the signal-to-noise ratio of the system. Combined with the advantages of high spatial resolution, high response bandwidth, high signal-to-noise ratio, and large strain range, the scheme has great potential in dynamic strain-sensing applications.
