Reliable sensing and accurate location of a weak and small hot spot are critical for applications in industrial infrastructure monitoring. We propose and experimentally demonstrate a practical and reliable distributed hot spot detection method using ultra-weak fiber Bragg gratings (UWFBGs) array and optical time-domain reflectometry (OTDR) based interrogator. To reliably detect the hot spots, the grating spacing of the sensor array is decreased to a similar size of the hot spot. All UWFBGs within a fiber section (FS) are considered as one sensing element, and the wavelength-division multiplexing technique is introduced to reduce crosstalk between adjacent FSs. To retrieve the sensing information, the proposed FS spectrum interrogation method based on OTDR technology is numerically analyzed and experimentally demonstrated. The interrogator exploits the reflection spectrum of each FS instead of each grating, enabling the low-speed hardware implementation of the whole demodulation method. Experimental results show that the expected hot spot can be successfully detected with a sensing resolution of 10 cm and a location resolution of 1 m over a range of 2 km by exploiting 10-ns pulsewidth. Besides, the temperature measurement can be demonstrated with a temperature sensing precision of ± 1°C and a measurement time of 1.5 s, which are meaningful for the early warning of centimeters-sized fire source in some oil and gas pipelines monitoring applications.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Raman distributed temperature sensing (RDTS) is a well-established optical fiber sensing technology for temperature sensing, which has been commercially available nowadays [1,2]. By using an optimized conventional RDTS scheme, the sensing range can reach several kilometers. With the help of advanced techniques such as optical pulse coding and optical amplification, this range can be extended up to a few tens of kilometers [3,4]. Apart from an extended sensing range, the spatial resolution of a temperature sensing system is also of great importance for reliable detection of small hot spots in applications such as downhole oil/gas production, structure health monitoring, fire prediction, among others [5–8]. To improve the spatial resolution of RDTS, researchers proposed several novel techniques. Liu et al. designed and fabricated a specific graded-index few-mode fiber to achieve a spatial resolution of 1.13 m . Scott et al. combined a pulse coding technique with RDTS, achieving 1-m spatial resolution over a 26-km fiber . Michael G. Tanner et al. demonstrated a RDTS with spatial resolution on the order of 1 cm using superconducting nanowire single-photon detectors . However, the fiber under test is only 2.5 m, and the implementation of such photodetectors in practical sensing applications is still nonviable. One drawback of RDTS is that the backscattered signal is typically weak. Therefore, thousands of averaging are needed to improve the signal-to-noise ratio (SNR), resulting in a long response time. Additionally, the temperature sensed by the RDTS technique is an averaged value over the entire fiber section of length equal to the spatial resolution. The obtained temperature is thus significantly underestimated when the size of the hot spot is small. Consequently, pre-warning of hot spots and fires would be delayed, leaving less time for the fire department to react.
Compared with the RDTS, FBG-based sensing exhibits much higher SNR [12,13] while providing the exact temperature information of the discrete points. With the development of an on-line FBG writing technique , the multiplexing capacity of an FBG array can be significantly increased using wavelength-division multiplexing (WDM), time-division multiplexing (TDM), or the combination of the two techniques [15–17]. The spatial resolution of FBG-based sensing depends on the grating spacing of the array. Combining with the optical frequency-domain reflectometry (OFDR) technique based demodulation, sub-millimeter spatial resolution can be achieved . In hot spot detection, however, demodulation based on the optical time-domain reflectometry (OTDR) technique is favored over OFDR due to the long sensing range with large multiplexing capacity . With OTDR, the pulsewidth of the optical source should match the spatial resolution in order to avoid signal crosstalk . With 20-ns pulsewidth, a spatial resolution of 2 m is realized based on a UWFBG array containing 2000 UWFBGs, with an expected interrogation speed up to 5000 FBGs per second . To further improve the spatial resolution, optical sources with even shorter pulsewidth need to be adopted, which unavoidably complicates the system setup .
In this paper, a large-capacity densely spaced UWFBG sensing array containing 20000 sensors with 10-cm grating spacing is employed to realize a highly reliable hot spot detection system. The sensor array is divided into 2000 one-meter fiber sections (FSs), each containing 10 UWFBGs. To reduce crosstalk, the center wavelengths of the FSs at 25 °C are designed to alternate between 1550.5 nm and 1552.5 nm. In contrast to traditional FBG-based sensing, the proposed system exploits the overall reflection spectrum of each FS instead of each grating because heating on any UWFBG in the FS results in a change of its overall spectrum. An FS spectrum interrogation method is proposed for the wavelength demodulation, which is able to recognize a hot spot affecting only one of the UWFBGs in an FS. Consequently, with the pulsewidth matching the 1-m FS length, the sensing resolution of the hot spot detection system is enhanced to 10 cm, with a location accuracy of 1 m. Due to the relaxed requirement on the pulsewidth, the hardware implementation of pulse modulation, signal detection, and data acquisition becomes much less costly. Temperature sensing precision of ± 1°C is achieved over a range of 2 km, consuming 1.5-s measurement time. The proposed sensing system is sensitive to hot spots like traditional FBG-based sensors, presenting no sensitivity-reducing averaging effect like the RDTS systems. It serves as a promising solution that provides a fast response with high sensing resolution and suitable location accuracy for hot spot detection.
2. Operation principle
Figure 1 depicts the large-capacity UWFBG array, which is divided into 2N FSs of length L, each containing m UWFBGs with uniform characteristics. The center wavelengths of the FSs are designed to alternate between λ1 and λ2, as shown in Fig. 1, in order to reduce signal crosstalk. Before the onset of a fire, the size of the hot spot can be only a few centimeters. To catch such a small hot spot, the grating spacing ΔL of the sensor array should be of similar size. In traditional FBG-based sensing systems adopting OTDR for wavelength demodulation, the pulsewidth of the optical sources needs to satisfy τ ≤ 2ΔLneff/c, where neff is the effective refractive index of the fiber and c is the speed of light in vacuum. Considering ΔL = 10 cm, the required pulsewidth τ should be no more than ∼ 1 ns. In that case, the optical transmitter and receiver, as well as the data acquisition system, would all need to ensure the bandwidths of several gigahertz. Instead of focusing on the reflection spectrum of a single UWFBG, we propose to exploit the overall reflection spectrum of the FS, which contains m gratings. Such a method is applicable because heating on a single UWFBG leads to a detectable change in the spectral shape of the overall FS spectrum. For a proper m, the requirement on the pulsewidth is thus relaxed to τ ≤ 2mΔLneff/c, which is much less stringent than that required by the traditional method. In the following, the FS spectrum interrogation method is explained in detail.
Based on the OTDR technique, the overall reflection spectrum of the FSs can be reconstructed. Because the two groups of FSs have different center wavelengths, they can be divided into two channels during the reconstruction, as shown by the purple part and the blue part in Fig. 2(a). Since hot spots cause the temperature to increase, the identification of spectral shape change and the quantization of the temperature rise is illustrated in Fig. 2(b). The blue curve illustrates the overall reflection spectrum of an FS at room temperature Tr. Considering a small hot spot increasing the temperature at only one of the UWFBGs in the FS to T0, as indicated by the small fire near FBG#n in Fig. 1, the reflection spectrum of FBG#n shifts towards the long-wavelength side, becoming a sidelobe of the main peak, as shown by the red curve. By setting a proper threshold, cross points between the threshold line and the spectra can be identified. The cross-point wavelength λr of the blue curve (corresponding to temperature Tr) is taken as the reference to obtain the wavelength shift. Identifying the cross-point wavelength λ0 of the red curve (corresponding to one UWFBGs in the FS with increased temperature T0) and taking the difference between λ0 and λr, the wavelength shift Δλ0 is measured. As the temperature rises further to T1, the sidelobe shifts further away, as shown by the green curve. The measured wavelength shift Δλ1 is also increased as compared to Δλ0, making the wavelength shift Δλ a suitable measurand for the quantification of the temperature. Considering a large hot spot increasing the temperature of all the UWFBGs in the FS to the same temperature T0, the FS spectrum also shifts towards the long-wavelength side, as shown by the purple curve, and the behavior is the same as that of a single grating. Compared to the red curve, the cross-point wavelength of the purple curve also follows the trend of the temperature change, only presenting a larger wavelength shift while the center wavelength is shifted for the same amount.
Further, the relationship between the measured wavelength shift Δλm and the actual wavelength shift Δλa (and thus the temperature change) is numerically analyzed. Ten apodized uniform FBGs with about 0.2-nm bandwidth, 20-dB side-mode suppression ratio, and 1550-nm center wavelength are modeled using the transfer matrix approximation method [21,22]. By assuming that the wavelength of one FBG (FBG#1) changes from 1550 to 1550.5 nm with an increment of 0.05 nm, the FS spectra with FBG#1 at different wavelengths are plotted in Fig. 3(a). With the wavelength of FBG#1 increasing, the reflection spectrum of FBG#1 shifts gradually towards the long-wavelength side and finally becomes isolated from the main peak, making it possible to detect the variation. Meanwhile, the intensity of the main peak gradually decreases to about 90% of the original peak value. The original FS spectrum, as shown by the black curve, is then taken as the reference, based on which the cross-point wavelength is obtained for each increment with different thresholds (Ith = 0.01, 0.02, 0.05, 0.08, 0.12), as shown in Fig. 3(b). When Ith = 0.01, as shown in the inset of Fig. 3(a), the position of the cross point is affected by the sidelobe of the FBG spectrum, which presents a relative peak intensity of 0.0108. Consequently, the results of the first two increments decrease as compared to the reference value, as shown by the black line in Fig. 3(b). Apart from these two points, the result shows an upward trend as the assumed center wavelength of FBG#1 increases. As the threshold increases, the measurement becomes less affected by the sidelobe. As shown by the red, the blue, and the magenta lines, which correspond respectively to Ith = 0.02, 0.05, 0.08. While the measured values are always smaller than the assumed wavelengths, the results are closer to the assumed values at lower Ith, indicating a higher temperature sensitivity in practice. When the threshold value is larger than the intensity of one single FBG (Ith = 0.12), we can see that the green line in Fig. 3(b) could not represent the assumed wavelengths. Accordingly, for better identification of the wavelength shifts, the threshold value slightly higher than the peak of the sidelobe should be selected.
By setting a proper threshold, the rightmost cross-point wavelength can be enough taken as the basis for detecting hot spots. However, there is a dead zone where the measured values are less sensitive to a small wavelength shift of one FBG, affecting the linearity of the measured values and the responsiveness of a small wavelength shift. Furthermore, the scenes where the spectra of multiple UWFBGs shift to the same wavelength are considered, as shown in Fig. 3(c). The intensity of the main peak decreases as the numbers of UWFBGs (NFBG) increase, while the intensity of the sidelobe of the FS spectrum increases, which makes it easier to detect hot spots. Setting Ith = 0.02, the differences (Dam) between Δλa and Δλm as the spectra of different numbers of UWFBGs shifted to different wavelengths are shown in Fig. 3(d). When NFBG = 1, the Dam varies between 30 and 50 pm as the wavelength shifts toward 1550.20 nm and then stabilizes at 50 pm as the wavelength continues to shift toward the long-wavelength side. When NFBG = 2, by contrast, the maximum of Dam decreases from 50 to 30 pm, and the variation range decreases from 20 to 12 pm, which means that the measured values are closer to the assumed values with a smaller fluctuation. Continue to increase the numbers of UWFBGs, the maximum and the variation range of Dam further decrease. When NFBG = 10, even the smallest assumed wavelength can be obtained, and the Dam is almost nonexistent. In practice, spectra of multiple UWFBGs may shift to different degrees due to the influence of the size and temperature of the hot spot. The temperature coefficients of the FS spectrum should adopt the same value for different situations because the number of FBGs affected by the hot spot is unknown. Taking a temperature coefficient of 10 pm/°C as an example, the measured temperature of one small hot spot will be about 5°C lower than its actual value. Nevertheless, compared to the RDTS, no matter what kinds of fire alarm strategies are chosen, the proposed interrogating method possesses a highly reliable response to small hot spots, which is meaningful for hot spot detection in oil and gas pipeline monitoring applications.
3. Experiment and results
3.1 Experimental setup
The interrogation system is depicted in Fig. 4. A tunable laser source (TLS) is adopted as the light source for 10-ns optical pulse generation using a semiconductor optical amplifier (SOA). An isolator is used to prevent the light from feeding back into the light source. After being amplified by an erbium-doped fiber amplifier (EDFA) and filtered by a 6-nm bandpass filter centered at 1552 nm, the pulsed light is launched into a sensing fiber containing 2000 FSs via a circulator. Each FS is composed of 10 UWFBGs, which are written directly on the fiber while it is drawn from the tower . The UWFBGs exhibit around − 43-dB reflectivity, 0.2-nm reflection bandwidth, 15-dB side-mode suppression ratio, and small grating spacing of 10 cm. To reduce the crosstalk between adjacent FSs, the center wavelengths are controlled to alternate between 1550.5 and 1552.5 nm. The sensing signals reflected by the FSs are converted to electrical signals by a high-performance photodetector (PD) with a bandwidth of about 80 MHz. The signals from the PD are digitized by a 500-MSPS analog-to-digital converter (ADC), and subsequently acquired by a field-programmable gate array (FPGA). The useful sensing information extracted from the acquired data by the FPGA is sent to a personal computer (PC) for further processing. Meanwhile, the FPGA synchronously controls the wavelength of the TLS, the pulse modulation by the SOA, and the optical gain of the EDFA. By scanning the wavelength of the TLS from 1550 to 1554 nm, the overall spectrum of each FS can be reconstructed from the reflected pulse trains, from which the wavelength information can be interrogated. With this setup, one round interrogation is completed within 1.5 s, which is mainly limited by the wavelength setting time of the laser.
3.2 Verification of the demodulation method
A 1-m FS containing 10 uniform UWFBGs is used to test the interrogation system. Different numbers of UWFBGs are placed into an environmental chamber with controlled temperature in order to mimic hot spots of different sizes. The temperature is varied from 25°C to 69°C, with an increment of 2°C. After each increment, we wait for 30 minutes before the measurement, so that thermal equilibrium is reached in the chamber. The rest UWFBGs are placed outside at room temperature of about 25°C.
Figures 5(a)−5(d) plot the overall reflection spectra of the FS at different temperatures and with different numbers of UWFBGs (NFBG = 1, 3, 5, 10) experiencing the temperature increase. Considering only one UWFBG is heated, as shown in Fig. 5(a), the spectrum of the specific UWFBG shifts out gradually as the temperature rises, presenting as a sidelobe next to the main peak. As the number of the heated UWFBGs increases to 3 and 5, the sidelobe becomes stronger, making it easier to detect, as shown in Figs. 5(b) and 5(c), respectively. When all the UWFBGs in the FS are heated simultaneously, the overall spectrum shifts with the temperature change, exhibiting the same behavior as a single grating. These experimental results further support the idea of the interrogation method described in Section 2.
We then verify how the threshold value affects the wavelength demodulation when only one UWFBG in the FS is affected by the hot spot. To seek out the appropriate threshold values, the SNR and the side-mode suppression ratio of each reconstructed spectrum are obtained firstly, which are 18 dB and 15 dB, respectively. Therefore, the threshold values Ith = 0.03, 0.04, 0.06, and 0.08 are tested for the wavelength demodulation of the FS spectra in Fig. 5(a). The wavelength shifts obtained at different temperatures with different threshold values are plotted in Fig. 6(a). When Ith = 0.03, the cross point of the threshold line and the spectrum is identified incorrectly due to the sidelobe of the FS spectrum (15dB, corresponding to 0.0316), which significantly affects the demodulation, as shown by the black curve. The demodulation results of the first 4 points barely increase, and others slightly grow, resulting in an R-square of 0.9534. When the threshold values are higher than the amplitude of the sidelobe, as shown by the red, blue, and purple curve, the demodulated shifts are much larger than that of the black curve, indicating significantly improved sensitivity. Meanwhile, the smaller the threshold values we choose, the more sensitive and linear the system is. When Ith = 0.04, the R-square of the results is improved to 0.9952, and the wavelength shifts at 69°C increased to 356.6 pm compared to that of 314.25 pm when Ith = 0.08. Besides, it is worth noting that there is a dead zone of 2°C where the system is less sensitive to the temperature, which is mainly caused by the sidelobe of the UWFBGs, as discussed in Section 2. The smaller the dead zone is, the more reliable the system is for detecting hot spots. Therefore, UWFBGs with high side-mode suppression ratios are favored in this application, which should be fabricated by the apodized method.
Considering the situation of more than one hot spot or a bigger hot spot in the same FS, Fig. 6(b) represents the relationship between wavelength shifts and temperatures with Ith = 0.04 when different numbers of FBGs experience the temperature rise. It is observed that the demodulated wavelength shifts are more sensitive to the temperature when more UWFBGs are heated. The wavelength shifts at 69°C increased to 435 pm when NFBG = 10 as compared to that of 356.6 pm when NFBG = 1. Meanwhile, due to the shrinking of the dead zones, better linearities are obtained between the temperature and the wavelength shifts. The disagreement between the experimental results and the simulated results in Fig. 3(d) comes from the spectral broadening caused by the slightly varying center wavelengths of the UWFBGs, which fluctuate less than 80 pm in an FS .
3.3 Distributed hot spot sensing
The interrogation method is then adopted for a quasi-distributed hot spot sensing based on the experimental setup in Fig. 4. By sending a pulse at a certain wavelength into the FS array, the serially reflected pulses from the FSs are collected, whose powers represent the reflectivities of the FSs at the corresponding wavelength. Figure 7(a) represents the reflected signals from the UWFBG array when the lasing wavelengths are 1550.5 and 1552.5 nm, respectively. According to the OTDR principle, different FSs can be differentiated by the time delays of the reflected sensing signals, and there are 2000 peaks (1000 FSs per wavelength) corresponding to the 2000 FSs in the sensing fiber. Due to the large multiplexing capacity, the reflected signals drop rapidly from around 0.6 to 0.2 V within a 2 km range, which is in accordance with the theoretical analysis. Compared with the front-end signal, the tail-end signal exhibits a slightly higher noise floor, which is caused mainly by the multiple-reflection effect among the UWFBGs . Figure 7(b) is a zoom-in view corresponding to the reflected signals in Fig. 7(a), and the amplitudes of the peaks correspond to the reflectivities of the FSs at the current wavelength. By scanning the lasing wavelength, the reflection spectra of FSs in the sensing fiber can be resolved. The reason why the reflected intensities of each FS are fluctuant is attributed to the small error in the fabricated parameters of individual gratings. Four 1550.5-nm FS spectra at different positions are plotted in Fig. 7(c). It can be seen that the spectral shadowing effect gradually appears as the distance increases (i.e., the number of UWFBGs increases), leading to spectral shape distortion in the center part of the reflection spectrum. Similarly, multiple-reflection crosstalk is more pronounced in the center part too. However, in the proposed wavelength interrogation method, the edge of the FS reflection spectrum is used, which is less affected by the spectral shadowing and the multiple-reflection crosstalk, especially when the reflection spectrum of the heat FBG shifts out of the FS spectrum. Consequently, the influences of these two effects, which limits the multiplexing capacity of sensors in the traditional FBG-based sensing system, is less detrimental in the proposed method.
To evaluate the demodulation accuracy of the system, two UWFBGs at the front end and the tail end of the array are heated by a semiconductor Peltier heater, while the rest of the UWFBGs are kept at 25°C. The heater temperature is increased from 25°C to 70°C with an increment of 5°C. Each temperature is kept unchanged for 30 minutes to ensure a uniform temperature distribution. Figure 8(a) presents the demodulation results of all sections under different temperatures. It is seen that the wavelength shifts of FS#1 and FS#1997 increase as the temperature increases, whereas the other FSs have no distinct wavelength shifts. Figure 8(b) shows the relationship between the wavelength shifts and the temperatures of these two FSs. The blue squares and the red dots correspond to the demodulation results of FS#1 and FS#1997 at each temperature. The blue and red curves correspond to the linear fitting results of FS#1 and FS#1997, respectively. The interrogation linearity of the blue curve is more than 0.9935, and the temperature coefficient is about 9.0586 pm/°C. In contract, the interrogation linearity of the red curve is reduced to 0.9881 due to the multiple-reflection effect , and the temperature coefficient is 8.9742 pm/°C. The temperature coefficients of the two FSs are both lower than that of a single grating (10 pm/°C), which is caused by the numbers of heated UWFBGs, as explained in Section 2. Setting the temperature coefficient to 10 pm/°C, the measurements at 70°C turn to be 65°C, which is consistent with the numerical analysis.
To evaluate the system stability, the wavelength demodulation experiment of the whole sensing fiber is repeated 30 times, with each round of interrogation completed in around 1.5 s. Figure 9 shows the measured wavelength shifts of all the FSs. It is observed that the error increases with the distance, showing a maximum error value is 9.56 pm at FS#1801. The enlarged error is mainly caused by SNR degradation, which is attributed to the multiple-reflection effect among the UWFBGs [18,24]. Besides, the wavelength shifts have a slight variation ranged from ± 3 to ± 10 pm, indicating that the precision of this system for temperature measurement is around ± 1°C.
In order to analyze the effectiveness of the proposed sensing technique, we compare the hot spot detection performance of the proposed system with a commercial Raman-based DTS and commercial linear heat detector. As shown in Fig. 10(a), three kinds of cables, including the UWFBG array cable, the optical fiber cable, and the heat-sensitive cable, are installed on the test bench side by side. A piece of heating tape with 10-cm width is placed above the cables to simulate a hot spot. The temperature of the heating tape is set at 50°C, while the temperature values of the two optical systems and the alarm messages of the electrical system are recorded. Figure 10(b) shows the temperature-time response curves of these three systems detecting the same location during the heating process. The blue, red, and purple lines correspond to the responses of the proposed system, the Raman DTS, and the electrical temperature sensor. As shown by the purple line, the electrical system gives an alarm approximately 70 s after the heating starts but does not provide detailed information on the temperature evolution. Meanwhile, an alarm from the electrical system only indicates the existence of a hot spot without any position information, which is disadvantageous as compared to the optical fiber sensing systems. The Raman DTS system provides time-resolved temperature and the corresponding position information. 140 s after the heating, the temperature measured by the Raman DTS eventually approaches 30°C, which is lower than the set temperature due to the inherent drawbacks, indicating a temperature-increase rate of around 2.57 °C/min. Taking an arbitrary week, the temperature could have a variation of almost 1 °C/min , and considering a very realistic scenario, 2.57 °C/min barely satisfies demand. With the proposed UWFBG array and the corresponding interrogation method, the detected temperature starts to increase 5 s after the heating starts. 140 s after the heating starts, the measured temperature has increased to 46°C. With the temperature-increase rate of around 9.43 °C/min, the system is suitable for hot spot detection in most industrial applications.
A practical, simple, and reliable distributed UWFBG sensing approach is introduced for small hot spot detection in this paper. To enhance the reliability, the grating spacing of adjacent UWFBGs is decreased to centimeter-level. All UWFBGs within each FS are considered as one sensing element, and the WDM technique is employed to reduce the crosstalk between adjacent FSs. The proposed FS spectrum interrogation method is numerically simulated and experimentally demonstrated for the wavelength demodulation, which is able to recognize a hot spot affecting only one of the UWFBGs in an FS. In contrast to the traditional FBG-based sensing, the proposed system exploits the overall reflection spectrum of each FS, instead of each grating, the hardware implementation of pulse modulation, signal detection, and data acquisition becomes much easier and less costly due to the relaxed requirement on the pulsewidth, which only needs to match with the length of each FS. Experimental results show that hot spot detection and location can be demonstrated with a 10-cm sensing resolution and a 1-m location resolution by using 10-ns pulsewidth. Besides, temperature sensing precision of ± 1°C is achieved over a range of 2 km, consuming 1.5-s measurement time. Furthermore, a comparison experiment with other kinds of DTS systems indicates that the proposed system exhibits great potential for the applications in fire detection and prevention.
National Natural Science Foundation of China (61705169, 61735013); Natural Science Foundation of Hubei Province (2018CFA056); China Scholarship Council (201906950046).
Our acknowledgments go to Eng. Ke Song and Eng. Kai Li from Wuhan FLOES Co., Ltd. for their precious help with the technical support of the hardware. The authors would also like to acknowledge Prof. Huiyong Guo and Dr. Yandong Pang in the National Engineering Laboratory for Fiber Optic Sensing Technology for their valuable discussions on paper writing.
The authors declare no conflicts of interest.
1. A. Ukil, H. Braendle, and P. Krippner, “Distributed Temperature Sensing: Review of Technology and Applications,” IEEE Sens. J. 12(5), 885–892 (2012). [CrossRef]
2. J. Li, B. Yan, M. Zhang, J. Zhang, B. Jin, Y. Wang, and D. Wang, “Long-Range Raman Distributed Fiber Temperature Sensor With Early Warning Model for Fire Detection and Prevention,” IEEE Sens. J. 19(10), 3711–3717 (2019). [CrossRef]
3. T. Lauber, G. Cedilnik, and G. Lees, “Physical Limits of Raman Distributed Temperature Sensing - Are We There Yet?” in 26th International Conference on Optical Fiber Sensors (OSA, 2018), p. WF30.
4. C. Yang, M. Wang, M. Tang, H. Wu, C. Zhao, T. Liu, S. Fu, and W. Tong, “Link optimized few-mode fiber Raman distributed temperature sensors,” Appl. Opt. 57(24), 6923–6926 (2018). [CrossRef]
5. Y. Dong, H. Zhang, L. Chen, and X. Bao, “2 cm spatial-resolution and 2 km range Brillouin optical fiber sensor using a transient differential pulse pair,” Appl. Opt. 51(9), 1229–1235 (2012). [CrossRef]
6. D. Kweon, K. Koo, J. Woo, and Y. Kim, “Hot spot temperature for 154 kV transformer filled with mineral oil and natural ester fluid,” IEEE Trans. Dielectr. Electr. Insul. 19(3), 1013–1020 (2012). [CrossRef]
7. Y. Bao, Y. Huang, M. Hoehler, and G. Chen, “Review of Fiber Optic Sensors for Structural Fire Engineering,” Sensors 19(4), 877 (2019). [CrossRef]
8. L. Hua, Y. Song, B. Cheng, W. Zhu, Q. Zhang, and H. Xiao, “Coherence-length-gated distributed optical fiber sensing based on microwave-photonic interferometry,” Opt. Express 25(25), 31362–31376 (2017). [CrossRef]
9. Y. Liu, L. Ma, C. Yang, W. Tong, and Z. He, “Long-range Raman distributed temperature sensor with high spatial and temperature resolution using graded-index few-mode fiber,” Opt. Express 26(16), 20562–20571 (2018). [CrossRef]
10. M. A. Soto, T. Nannipieri, A. Signorini, A. Lazzeri, F. Baronti, R. Roncella, G. Bolognini, and F. D. Pasquale, “Raman-based distributed temperature sensor with 1 m spatial resolution over 26 km SMF using low-repetition-rate cyclic pulse coding,” Opt. Lett. 36(13), 2557–2559 (2011). [CrossRef]
11. M. G. Tanner, S. D. Dyer, B. Baek, R. H. Hadfield, and S. Woo Nam, “High-resolution single-mode fiber-optic distributed Raman sensor for absolute temperature measurement using superconducting nanowire single-photon detectors,” Appl. Phys. Lett. 99(20), 201110 (2011). [CrossRef]
12. R. Cheng, L. Xia, C. Sima, Y. Ran, J. Rohollahnejad, J. Zhou, Y. Wen, and C. Yu, “Ultra-short FBG based distributed sensing using shifted optical Gaussian filters and microwave-network analysis,” Opt. Express 24(3), 2466–2484 (2016). [CrossRef]
13. J. Hervás, D. Barrera, J. Madrigal, and S. Sales, “Microwave Photonics Filtering Interrogation Technique Under Coherent Regime For Hot Spot Detection on a Weak FBGs Array,” J. Lightwave Technol. 36(4), 1039–1045 (2018). [CrossRef]
14. H. Guo, J. Tang, X. Li, Y. Zheng, H. Yu, and H. Yu, “On-line writing identical and weak fiber Bragg grating arrays,” Chin. Opt. Lett. 11(3), 030602 (2013). [CrossRef]
15. Z. Luo, H. Wen, H. Guo, and M. Yang, “A time- and wavelength-division multiplexing sensor network with ultra-weak fiber Bragg gratings,” Opt. Express 21(19), 22799–22807 (2013). [CrossRef]
16. Y. Wang, J. Gong, B. Dong, D. Y. Wang, T. J. Shillig, and A. Wang, “A Large Serial Time-Division Multiplexed Fiber Bragg Grating Sensor Network,” J. Lightwave Technol. 30(17), 2751–2756 (2012). [CrossRef]
17. Y. Ou, C. Zhou, L. Qian, D. Fan, C. Cheng, and H. Guo, “Large-capacity multiplexing of near-identical weak fiber Bragg gratings using frequency-shifted interferometry,” Opt. Express 23(24), 31484–31495 (2015). [CrossRef]
18. X. Gui, Z. Li, X. Fu, C. Wang, H. Wang, F. Wang, and X. Bao, “Large-scale multiplexing of a FBG array with randomly varied characteristic parameters for distributed sensing,” Opt. Lett. 43(21), 5259–5262 (2018). [CrossRef]
19. H. Guo, L. Qian, C. Zhou, Z. Zheng, Y. Yuan, R. Xu, and D. Jiang, “Crosstalk and Ghost Gratings in a Large-Scale Weak Fiber Bragg Grating Array,” J. Lightwave Technol. 35(10), 2032–2036 (2017). [CrossRef]
20. Y. Wang, J. Gong, D. Y. Wang, B. Dong, W. Bi, and A. Wang, “A Quasi-Distributed Sensing Network With Time-Division-Multiplexed Fiber Bragg Gratings,” IEEE Photonics Technol. Lett. 23(2), 70–72 (2011). [CrossRef]
21. K. S. Khalid, M. Zafrullah, S. M. Bilal, and M. A. Mirza, “Simulation and analysis of Gaussian apodized fiber Bragg grating strain sensor,” J. Opt. Technol. 79(10), 667 (2012). [CrossRef]
22. M. Prabhugoud and K. Peters, “Modified Transfer Matrix Formulation for Bragg Grating Strain Sensors,” J. Lightwave Technol. 22(10), 2302–2309 (2004). [CrossRef]
23. J. Wang, Z. Li, Q. Yang, X. Fu, X. Gui, C. Wang, and H. Wang, “Interrogation of a large-capacity densely spaced fiber Bragg grating array using chaos-based incoherent-optical frequency domain reflectometry,” Opt. Lett. 44(21), 5202–5205 (2019). [CrossRef]
24. H. Guo, F. Liu, Y. Yuan, H. Yu, and M. Yang, “Ultra-weak FBG and its refractive index distribution in the drawing optical fiber,” Opt. Express 23(4), 4829–4838 (2015). [CrossRef]
25. T. T. Aralt and A. R. Nilsen, “Automatic fire detection in road traffic tunnels,” Tunn. Undergr. Space Technol. 24(1), 75–83 (2009). [CrossRef]