Open Access
Add to BibSonomy
Abstract
Low noise, good consistency, and long-term stability are critical for a multi-channel fiber optic intrusion detection system. This paper proposes a low-noise 32-channel fiber optic interrogator based on phase generation carrier technique, which emphasizes on the analysis of key parameters related to system consistency and stability. A novel time-division locking technique of the carrier modulation depth and carrier phase delay is proposed. By locking the carrier modulation depth and carrier phase delay simultaneously, the consistency and stability of the multi-channel interferometric interrogator is significantly improved. The results show that in the multi-channel system, the noise level of each channel is below 12.6 ng/√Hz @ 5 Hz, and the noise level differences between each two channels are less than 2.6 dB, the noise level fluctuations of all channels are less than 1.4 dB over 24 hours. A subway intrusion detection is demonstrated in Shenzhen, China. Geological drilling events can be clearly recorded and identified.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Subway is one of the most expensive infrastructure and important lifeline project in the city. The safety of the metro rail needs to be guaranteed when it passes through the machinery construction area on the ground. However, illegal drilling above the subway tunnel seriously threatens the safety of the passing subway train. In the past three years, there have been several subway intrusion accidents caused by geological drill, one of the most common type of drilling machine, in China. Recently, engineering seismic methods have been used to detect and identify the vibration events above subway tunnels [1,2]. Optical fiber vibration sensors have obvious advantages in applications in subway tunnel due to their advantages such as anti-electromagnetic interference, long-distance transmission and distributed measurement. Nowadays, distributed optic fiber sensors based on phase-sensitive optical time domain reflectometer (ϕ-OTDR) and fiber Bragg grating (FBG) sensors are widely used for intrusion detection of urban underground structure [3,4]. However, the vibration signal caused by the geological drill on the ground is very weak when transmitted into the tunnel. The fiber optic sensing system located in the tunnel must have extremely low noise to record the vibration event. Compared with the two fiber optic sensors above, fiber optic interferometric seismometer has at least two to three orders of lower noise level, which makes it a distinct advantage in the recording of weak geological drilling intrusion events [5]. In addition, to accurately locate the drilling signals under different geological conditions, a sensor array is necessary. Each sensor in the array should have good consistency and long-term stability.
For fiber optic interferometric sensor, phase generation carrier (PGC) based interrogation technique is widely used for signal demodulation [6,7]. The consistency and stability of the interferometric fiber optic sensing system based on PGC method are the key problems that need to be solved in practical applications [8]. In the past, researchers have found that the carrier modulation depth and carrier phase delay are the two factors affecting the stability and consistency of the PGC system [9–16]. In particular, in order to ensure that the interferometric sensor has the best state of noise level and harmonic distortion, it is necessary to control the carrier modulation depth to a specific value. However, the carrier modulation depth is affected by the arm length difference of each fiber-optic interference sensor and the frequency shift of the modulated laser source. This is also affected by the ambient temperature fluctuation and the stability of the modulation voltage [9–12]. Therefore, for a multi-channel interferometric fiber-optic sensing system, the modulation depth of each sensor will be different between every two channels and will change at any time, which will affect the consistency and long-term stability of the system. On the other hand, due to the effects of digital-to-analog conversion delay, laser PZT response delay, optical transmission time, and photoelectric conversion delay, there is a certain phase delay between the interference fringe and the modulation carrier, which has a negative effect on the system's noise level and harmonic distortion, which in turn changes the consistency and stability of the system [13–16]. Obviously, in order to achieve a multi-channel optical fiber interferometric sensor system with high consistency and long-term stability, the carrier modulation depth and carrier phase delay of each channel need to be well controlled.
To solve these problems, the conventional approaches are to develop non-linear calibration compensation algorithms to compensate the effects of the fluctuations of carrier modulation depth and carrier phase shift delay [9,10,12,14–16]. But these methods are still affected by ambient temperature and laser frequency drift. These algorithms will also become very complicated in a multi-channel interferometric fiber sensing system, as each channel requires different calibration parameters. Especially these algorithms may even fail when the carrier modulation depth changes greatly. Recently, the real-time feedback techniques of modulation depth and phase carrier delay of a single-channel PGC interferometric sensing system are studied [11,13]. This can ensure that the single-channel fiber-optic interferometric fiber sensing system maintains high stability and low harmonic distortion. However, for a multi-channel PGC-based optic fiber interferometric sensing system, each interferometric sensor has different carrier modulation depth values and carrier phase carrier delays. Thus, simultaneous feedback locking of carrier modulation depth and carrier phase delay becomes critical to improve the consistency and long-term stability in a multi-channel interferometric sensing system.
This paper proposes a novel time-division locking technique of the carrier modulation depth and carrier phase delay to improve the consistency and stability of PGC-based fiber optic interferometric sensing system which is used for subway intrusion detection. The analysis of factors affecting system consistency and stability are given in details. The principle and test results of the time-division locking technique and the novel algorithm are highlighted. In particular, we demonstrated the multi-channel fiber optic intrusion monitoring system in Shenzhen subway based on the proposed interrogator and seismometer array. An experiment of geological drilling intrusion detection is carried out and the results and discussion are given.
2. Principle and analysis of multi-channel fiber optic interrogator
The 32-channel PGC-based interferometric fiber optic phase interrogator is shown in Fig. 1. A narrow linewidth fiber laser (NKT, X15), modulated by a digital to analog converter driven by a direct digital synthesis (DDS) with 5 kHz sine wave, is used for light source. A space division multiplexing technique using a 1×32 couple is used to extend the channels number of the system. The interferometric signals are received by photo-detectors (PDs) and amplifiers. The electrical output of the amplifier is digitalized in accordance with the carrier signal using an analog-to-digital convertor. The PGC phase demodulation is accomplished on a field programmable gate array (FPGA) board (NI PXIe 7858R).
Fig. 1. The basic schematic of the 32-channel PGC-based interferometric fiber optic phase interrogator. CP, coupler; ISO, isolator; CIR, circulator; PD, photodetector; DAC, digital to analog converter; DDS, direct digital synthesis; A/D, analog-to-digital converter; FPGA, field programmable gate array.
For a PGC-Arctan-based Michelson interferometric interrogation system, the calculated phase information can be expressed by the following formula when the carrier phase delay is considered [13].
We can find that the phase result to be demodulated is directly related to the carrier modulation depth and carrier phase delay. In order to ensure that the interferometric sensing system has the best state of noise level and harmonic distortion, it is necessary to control the carrier modulation depth to 2.63, the carrier phase delay to 0 rad. However, due to the effects of digital-to-analog conversion delay, laser PZT response delay, optical transmission time, and photoelectric conversion delay, there is a certain phase delay between the interference fringe and the modulation carrier, which has a negative effect on the system's noise level and harmonic distortion, which in turn changes the consistency and stability of the system [13–16]. The carrier modulation depth can be expressed by the formula (2) [17].
where, n is the refractive index of the fiber, c is the speed of light in a vacuum, Δl is the arm length difference between the two arms of the Michelson interferometer, and Δν represents the frequency shift of the modulated laser source.We can find that the carrier modulation depth is greatly affected by the arm length difference of each fiber-optic interference sensor and the modulation amplitude of laser light source frequency, which is related to the ambient temperature fluctuation and the stability of the modulation voltage [9–12]. Therefore, in order to achieve a multi-channel optical fiber interferometric sensor system with good consistency and long-term stability, the carrier modulation depth and carrier phase delay of each channel need to be well controlled.
3. Time-division locking technique for improving consistency and stability
3.1 Time-division locking technique
The basic principle of the proposed time-division locking technique is shown in Fig. 2. We divide the system into four modules, fiber optic interrogation and photoelectric detection module, algorithm module, analog-to-digital convertor (ADC) module and digital to analog convertor (DAC) module. The core idea of this technology is to make the phase measurement of each channel a closed-loop system according to time division, and realize the feedback lock of carrier modulation depth (C) and carrier phase delay (CPD) one by one. The carrier modulation depth and carrier phase delay are calculated by Refs. [11,13]. A Direct Digital Synthesis (DDS) is used to generate carrier signals with different amplitudes and initial phases for the fiber optic seismometer array. The carrier modulation depth and carrier phase delay of each channel seismometer can be controlled to be 2.63 and 0 rad simultaneously by changing the amplitudes and initial phases of DDS. Then the interference fringes of each channel are collected one by one. A phase measurement sequence is recognized by PGC-Arctan demodulation algorithm. In this process, the carrier modulation depth and carrier phase delay of each channel are effectively controlled by real-time feedback and are not affected by the external environment, which is conducive to improve the consistency and stability between each channel.
Fig. 2. The basic principle of the proposed time-division locking technique. Ci is the carrier modulation depth of the i-th channel. DDSwi is modulated carrier applied to the i-th channel. IFi is the interference fringes of the i-th channel. ADC is analog-to-digital convertor. DAC is digital to analog convertor.
To realize fast, continuous and simultaneous feedback lock of the modulation depth and phase delay of each channel, a real-time calculation and feedback control algorithm need to be designed, which is shown in Fig. 3. This algorithm contains eight parts. The modules of Acq., Bandpass Filter, DDS, and CPD and C calculation are used to obtain interference fringe signals and calculate the values of CPD and C. The module of CPD and C Feedback is used to calibrate modulation depth and carrier phase delay for each channel. The modules of PGC-Arctan and Entire Phase Space Expansion are used to calculate phase information for each channel. Then the multi-channel phase demodulation results are obtained by the module of Time-Division Multichannel Sampling. In order to ensure that the phase information of each channel is obtained under the stable of C and CPD. The feedback control of all channels is performed sequentially in terms of time. The total feedback time should match the sampling time in the feedback locking process. Thus, the sum of the feedback control time of all channels within one locking loop should be less than the data sampling rate of the interference fringes. As shown in Fig. 3, the coarse and fine feedback scheme is important to quickly and accurately realize the feedback lock of C and CPD of each sensor. In order to achieve accurate feedback control of C and CPD, the voltage amplitude and phase step values are set to be very small (0.001 V and 0.001 rad). However, taking the voltage amplitude drive as an example, the voltage amplitude may have a large numerical change (such as 0.1 V) when the system switches from the current lock state to the next state. It will take a long time (100 cycles) to adjust the driving voltage. For a 500 kHz clock, 100 cycles are equivalent to 0.2 milliseconds, so it takes at least 6.4 milliseconds to complete the drive voltage adjustment of 32-channel sensors (considering that the feedback lock requires a stabilization time, the time will be longer). In theory, for a 32-channel demodulation system with a sampling rate of 1 kS/s, the total feedback time of all sensors cannot exceed 1 millisecond. Therefore, it is necessary to make a coarse feedback before accurately adjusting the voltage drive. By recording the current drive voltage amplitudes of each sensors using the registers, the next drive voltage can be roughly adjusted to the desired value in only one cycle. It will only take a short time for fine feedback because the modulation depth does not change much in short term.
In order to further explain the process of time division locking, Fig. 4 shows the results of the time division locking of C and CPD of the two-channel sensing system. Figure 4(a) shows the changes of the interference fringes between the two channels during the feedback process due to changes in the voltage amplitude and initial phase of the DDS output. Figure 4(b) and Fig. 4(c) are the results of the amplitude adjustment changes of C and CPD of the two channels respectively. We can see that the C value is precisely controlled at 2.63 when the channel is locked. The CPD fluctuation is less than 0.003 rad. The sampling rate of the original interference fringes is set to 500 kS/s. Each channel actually spends 0.1 millisecond during the locking process. For the proposed 32-channel fiber optic interrogator, the feedback lock time of each sensor is controlled at 0.01 millisecond. The total feedback time of all sensors within one locked loop is 0.32 milliseconds, which is less than the single synchronization sampling time of a 32-channel sensor with a sampling rate of 1 kS/s. Therefore, for each acquisition cycle of original interference fringes, C and CPD of all channels are always controlled steadily. In other words, we can think that C and CPD are always continuously and simultaneously locked for the multi-channel sensing system. The feedback time determines the minimum sampling rate of original data and the number of channels that can be expanded. Higher sampling rate can provide more time for feedback control and help increase the number of channels.
Fig. 4. Time-division locking process: two typical interference fringes (a), the carrier modulation depth of the two channels (b), and the carrier phase delay of the wo channels (c).
3.2 Consistency and stability test
In order to measure the noise level of the multi-channel fiber optic interrogator, we place the fiber optic seismometers in a quiet and vibration-isolated environment. A 1 Hz high-pass filter is used to eliminate low-frequency noise effects. The noise level of the multi-channel interrogator is shown in Fig. 5. We can find that the noise levels of the three selected channels are in good agreement with a level of about 118 dB @ 5 Hz (1.26×10−6 rad/√ Hz @ 5 Hz). The system noise level is about 12.6 ng/√Hz @ 5 Hz considering the fiber optic seismometers have the sensitivity of 40 dB (re: 0 dB = 1 rad/g) [18].
To further test the consistency and long-term stability between channels of a multi-channel interrogator, the average background noise levels of all channels of the multi-channel interrogator from 1 Hz to 500 Hz are calculated in the laboratory within 24 hours, which is shown in Fig. 6. The temperature range is from 16 ℃ to 27 ℃. We can find that the noise level differences between every two channels are less than 2.6 dB, the noise level fluctuations of all channels are less than 1.4 dB over 24 hours.
Fig. 6. The fluctuation curves of the average background noise level of different channels of the multi-channel fiber optic seismometers within 24 hours.
4. Demonstration of subway intrusion detection
To evaluate the performance of the multi-channel interrogator, a demonstration of subway intrusion detection is carried out in Shenzhen, China. A fiber optic seismometer array is installed on the tunnel segment of the Liyumen subway station for recording the vibration signal caused by the geological drill above the tunnel. The array consists of 32 seismometers with a spacing of 5 meters to 25 meters. The fiber optic seismometer is based on clamped diaphragms. The phase-acceleration sensitivity of the fiber optic seismometer is calibrated by standard shake table. The fiber optic seismometers are selected to have a flat sensitivity response, and good sensitivity consistency. According to our previous report [18], the sensitivity of the seismometer is about 41 dB (re: 0 dB = 1 rad/g) with a fluctuation less than ± 1.5 dB within the frequency range from 1 Hz to 500 Hz. The sensitivity difference between the different sensors is less than ± 1.5 dB. The consistency and stability of the intrusion detection system will mainly be determined by the interrogator since the sensitivity of the fiber optic seismometer has good flatness and consistency.
The multi-channel interrogator is placed in the subway station monitor room. An automatic alarm software is developed for automatic identification of geological drilling and send alarm information via 4G network. Geological drill is carried out on the square above the Liyumen subway station shown in Fig. 7. In order to eliminate the influence of noise interference from trains, construction machinery on the ground, and blowers, we selected noise in quiet conditions to evaluate the stability of the system from 0 am to 4 am in the subway tunnel. The temperature range is from 22 ℃ to 25 ℃. As a result, the noise fluctuations of the system are less than 2 dB, which shows that the system still has good stability under temperature fluctuations. It should be noted that the average noise level and noise fluctuations of the system working on site are higher than that of the laboratory due to the higher background noise of the tunnel.
Fig. 7. The basic principle and scheme of the intrusion detection system based on the multi-channel fiber optic seismometer array.
Figure 8 shows the time domain waveform and frequency map before and after geological drilling. The test results show that the proposed system can clearly record the geological drilling construction signals. In Fig. 8 (a), the geological drill is located directly above the first channel sensor (CH 1). The channels from CH 1 to CH 7 represents the distance from near to far. We can see that the farther the distance is, the smaller the recorded acceleration amplitude (about 10−3 m/s2) of the fiber optic seismometers. From Fig. 8 (b) and (c), we can find that the environmental noise does not have a very large dominant frequency before the geological drilling operation. For subway tunnel intrusion monitoring, locating the geological drilling precisely is very important. The nearest sensor to the geological drilling can be identified according to the signal amplitude. If the geological drill signals are recorded by more than one sensor, we can easily locate the position of the geological drill roughly by the amplitude of the sensors. The locating accuracy is determined by the distance between the two adjacent sensors. In our configuration, as shown in Fig. 7, the minimum spacing in the optic fiber seismometer array is 5 meters. So a locating resolution of 5 meters along the axial direction of the tunnel can be achieved. The use of two optic fiber seismometer arrays, installed respectively in each tunnel in the two-way tunnel, are helpful to further improve the locating accuracy. This is also useful to locate the drill in the normal direction of the tunnel by comparing the signals from the two sensor arrays. In addition, we can also use geophysical exploration methods to help improve the locating accuracy by comparing the time difference of the vibration signals recorded by different sensors, which is an important work in the future.
Fig. 8. The time domain waveform of the recorded geological drill signals (a), the spectrograms of the recorded signals before (b) and after (c) geological drilling.
The vibration signal caused by the geological drilling obviously has a dominant frequency component of 41 Hz. During the test, we found that this dominant frequency will vary slightly with the different geological drilling conditions and operation status. We can identify the geological drilling events above the tunnel by the time domain characteristics and frequency characteristics of the recorded vibration signals. We can guarantee that the proposed system has a high identification rate for geological drilling signals due to the high consistency and long-term stability of the proposed time-division locking technique.
Conclusion
In this paper, we demonstrate a multi-channel fiber optic interferometric interrogator with low-noise level, good consistency, and long-term stability. A novel time-division locking technique of the carrier modulation depth and carrier phase delay is proposed in the multi-channel interrogator. Test results show that using the proposed time-division locking technique the system noise level is below 12.6 ng/√Hz @ 5 Hz, the noise level differences between each channel are less than 2.6 dB, the noise fluctuations of all channels are less than 1.4 dB over 24 hours. The demonstration in Shenzhen subway shows that geological drilling events can be clearly recorded and identified due to the highly consistent and stable time-frequency characteristics of the recordings. The result also implies the proposed time-division locking technique for multi-channel interrogator is promising in the early warning of subway tunnel intrusion.
Funding
National Key Research and Development Program of China (2018YFB2101003); National Natural Science Foundation of China (U1939207); The Scientific Instrument Developing Project of the Chinese Academy of Sciences (Broadband fiber optic seismometer acquisition instrument and system); Youth Innovation Promotion Association of the Chinese Academy of Sciences (2016106).
Acknowledgments
Thanks for the support and help from Shenzhen Emergency Management Bureau, Shenzhen Rail Transit Construction Headquarters Office and Shenzhen Metro Group. The authors also thank the help in the experiment from Yusong Shen, Jinzhu Su, Zhi Liu, Xiaoqing He, and Wenhui Huang from Guangdong Earthquake Administration.
Disclosures
The authors declare no conflicts of interest.
References
1. A. H. Peltz, J. L. Volterra, A. Delle, and F. Streichenwein, “Urban blasting vibrations: case histories of vibration monitoring in New York city,” International Conference on Case Histories in Geotechnical Engineering 17 (2008).
2. https://www.instantel.com/applications/tunnel-subway-monitoring
3. R. Peng, Z. Liu, and S. Li, “Perimeter monitoring of urban buried pipeline subject to third-party intrusion based on fiber optic sensing and convolutional neural network,” Proc. SPIE 11209, 112091Z (2019). [CrossRef]
4. W. Gan, S. Li, Z. Li, and L. Sun, “Identification of ground intrusion in underground structures based on distributed structural vibration detected by ultra-weak FBG sensing technology,” Sensors 19(9), 2160 (2019). [CrossRef]
5. J. Huang, W. Zhang, W. Huang, W. Huang, L. Wang, and Y. Luo, “High-resolution fiber optic seismic sensor array for intrusion detection of subway tunnel,” In Proceedings of the 2018 Asia Communications and Photonics Conference (ACP), IEEE1–3 (2018)
6. Q. Lin, L. Chen, S. Li, and X. Wu, “A high-resolution fiber optic accelerometer based on intracavity phase-generated carrier (PGC) modulation,” Meas. Sci. Technol. 22(1), 015303 (2011). [CrossRef]
7. A. D. Kersey, “Recent progress in interferometric fiber sensor technology,” Proc. SPIE 1367, 2–12 (1991). [CrossRef]
8. Q. Shi, L. Wang, H. Zhang, C. Tian, M. Zhang, Q. Tian, Y. Liao, H. Wang, X. Zeng, and L. Huang, “The stability and consistency analysis of optical seismometer system using phase generated carrier in field application,” Proc. SPIE 7508, 75081M (2009). [CrossRef]
9. Z. Qu, S. Guo, C. Hou, J. Yang, and L. Yuan, “Real-time self-calibration PGC-Arctan demodulation algorithm in fiber-optic interferometric sensors,” Opt. Express 27(16), 23593 (2019). [CrossRef]
10. S. Zhang, L. Yan, B. Chen, Z. Xu, and J. Xie, “Real-time phase delay compensation of PGC demodulation in sinusoidal phase-modulation interferometer for nanometer displacement measurement,” Opt. Express 26(2), 605–616 (2018). [CrossRef]
11. A. V. Volkov, M. Y. Plotnikov, M. V. Mekhrengin, G. P. Miroshnichenko, and A. S. Aleynik, “Phase modulation depth evaluation and correction technique for the PGC demodulation scheme in fiber-optic interferometric sensors,” IEEE Sens. J. 17(13), 4143–4150 (2017). [CrossRef]
12. L. Hou, F. Peng, J. Yang, Y. Yan, D. Yan, B. Wu, and L. Yuan, “An improved PGC demodulation method to extend dynamic range and compensate low-frequency drift of modulation depth,” Proc. SPIE 9634, 96345Y (2015). [CrossRef]
13. A. N. Nikitenko, M. Y. Plotnikov, A. V. Volkov, M. V. Mekhrengin, and A. Y. Kireenkov, “PGC-Atan demodulation scheme with the carrier phase delay compensation for fiber-optic interferometric sensors,” IEEE Sens. J. 18(5), 1985–1992 (2018). [CrossRef]
14. S.-C. Huang and H. Lin, “Modified phase-generated carrier demodulation compensated for the propagation delay of the fiber,” Appl. Opt. 46(31), 7594–7603 (2007). [CrossRef]
15. S. Zhang, B. Liping Yan, Z. X. Chen, and J. Xie, “Real-time phase delay compensation of PGC demodulation in sinusoidal phase-modulation interferometer for nanometer displacement measurement,” Opt. Express 25(1), 472–485 (2017). [CrossRef]
16. S. Li, S. Shao, H. Mei, Q. Hao, and R. Rao, “Analysis and mitigation of the carrier phase delay effect of the digital phase generated carrier algorithm,” Appl. Opt. 56(3), 731–738 (2017). [CrossRef]
17. A. Dandridge, A. B. Tveten, and T. G. Giallorenzi, “Homodyne Demodulation scheme for fiber optic sensors using phase generated carrier,” IEEE Trans. Microwave Theory Tech. 30(10), 1635–1641 (1982). [CrossRef]
18. D. Jiang, W. Zhang, and F. Li, “All-Metal optical fiber accelerometer with low-transverse sensitivity for seismic monitoring,” IEEE Sens. J. 13(11), 4556–4560 (2013). [CrossRef]
