Underground cable location using high-speed time series template matching
The method of fast time-series template matching with DFOS efficiently locates underground cables in real-time using unique vibration patterns, overcoming inefficiencies and errors in existing methods, and achieving rapid, accurate localization with minimal resources.
Patent Information
- Application Number
- JP2024525688
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2022-11-02
- Filing Date
- 2022-11-03
- Publication Date
- 2024-11-06
- Estimated Expiration
- 2042-11-03
AI Technical Summary
Existing methods for locating underground cables are inefficient, time-consuming, and prone to errors, especially when integrating communication and sensing systems over a common fiber, and require extensive training data for neural networks.
A method using fast time-series template matching with distributed fiber optic sensing (DFOS) that employs unique vibration patterns distinguishable from background noise, allowing real-time detection of cable locations using portable devices without the need for training data, and optimized for resource-constrained platforms.
Enables rapid and accurate localization of underground cables with low false alarms, reducing operational time and equipment weight, and providing real-time feedback using lightweight, portable devices.
Smart Images

Figure 2024541046000001_ABST
Abstract
Description
[Technical field]
[0001] The present disclosure relates generally to distributed fiber optic sensing (DFOS) systems, methods, and structures, and more particularly to underground cable localization using fast time series template matching combined with DFOS. [Background technology]
[0002] A significant challenge facing communications service providers worldwide is efficiently managing the millions of miles of buried and suspended optical fibers that comprise their network infrastructure. Accordingly, systems, methods, and structures that facilitate locating buried and suspended optical fibers would be a welcome addition to the art. Summary of the Invention
[0003] An advancement in the art is made by aspects of the present disclosure that relate to distributed fiber optic sensing underground cable localization using high speed time series data template matching.
[0004] In contrast to the prior art, the system and method according to the present invention provides cable localization based on time series similarity search. The present technology can be used to rapidly search either specially designed vibration patterns created with on / off patterns distinguishable from background noise, or real field vibration signals collected from secondary point sensors (e.g., cell phones, tablets, accelerometers, etc.). Furthermore, the present system and method works with different signal patterns generated by different vibrators, of different durations, and in different locations (e.g., buried or aerial portions of the cable), and uses resource-limited platforms (e.g., laptops and edge devices) so that field technicians do not have to wait more than a few seconds to receive results.
[0005] A method for underground cable location using fast time series template matching and distributed optical fiber sensing (DFOS) includes providing a DFOS system including an optical sensor fiber, a DFOS interrogator in optical communication with the optical sensor fiber, the DFOS interrogator configured to generate optical pulses, couple the generated pulses into the optical sensor fiber, and receive backscattered signals from the optical sensor fiber, and an intelligent analyzer configured to analyze DFOS data received by the DFOS interrogator and determine vibration activity occurring at a location along the optical sensor fiber from the backscattered signals, disposing a programmable vibration generator at a field location proximate to the optical sensor fiber, transmitting to the programmable vibration generator a unique vibration pattern generated by the vibration generator, activating the programmable vibration generator to generate the transmitted unique vibration pattern, activating the DFOS system and collecting / analyzing the determined vibration activity to further determine vibration activity indicative of the unique vibration pattern generated by the vibration generator. [Brief description of the drawings]
[0006] A more complete understanding of the present disclosure may be realized by reference to the accompanying drawings.
[0007] [Figure 1(A)] FIG. 1 is a schematic diagram illustrating a DFOS system according to an embodiment of the present disclosure.
[0008] [Figure 1(B)] FIG. 1 is a schematic diagram illustrating a coded constant amplitude DFOS system with out-of-band signal generation, according to an aspect of the present disclosure.
[0009] [Diagram 2] FIG. 1 is a schematic diagram outlining a method according to an aspect of the present disclosure.
[0010] [Diagram 3]1 is a graph of an exemplary periodic square wave signal with a 4 second period having on / off cycles of equal length with a sampling rate of 100 Hz, in accordance with an embodiment of the present disclosure.
[0011] [Figure 4] 1 is a graph of an exemplary periodic square wave signal with a 4 second period having equal length on / off cycles with a sampling rate of 100 Hz, in accordance with an embodiment of the present disclosure.
[0012] [Figure 5(A)] 1 is a pair of graphs of amplitude versus time of an actual vibration signal having a specified pattern overlaid after Z-normalization processing, in accordance with an embodiment of the present disclosure. [Figure 5(B)] 1 is a pair of graphs of amplitude versus time of an actual vibration signal having a specified pattern overlaid after Z-normalization processing, in accordance with an embodiment of the present disclosure.
[0013] [Figure 6] FIG. 13 illustrates an example sliding window that iterates through all n-m+1 sliding windows in ts to find the sliding window position that provides the maximum correlation to a query, according to an aspect of the disclosure.
[0014] [Figure 7(A)] 1 is a pair of graphs of amplitude versus time of actual vibration signals from a waterfall image, where the vertical lines are the start and end positions of the sliding window in which the subsequence is most similar to the query, according to an aspect of the present disclosure. [Figure 7(B)] 1 is a pair of graphs of amplitude versus time of an actual vibration signal from a waterfall image showing an overlay of both a query and a subsequence, in accordance with aspects of the present disclosure.
[0015] [Figure 8] FIG. 2 is an exemplary flow diagram illustrating tuple iteration in accordance with an aspect of the present invention.
[0016] [Figure 9] FIG. 1 is an exemplary flow diagram illustrating the overall operation of systems and methods according to aspects of the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0017] The following merely illustrates the principles of the present disclosure, and it will thus be appreciated that those skilled in the art will be able to devise various arrangements which, although not explicitly described or shown herein, embody the principles of the present disclosure and are included within its spirit and scope.
[0018] Furthermore, all examples and conditional terms described herein are intended to be for educational purposes only to aid the reader in understanding the concepts contributed by the inventors to further the principles and techniques of the present disclosure, and should not be construed as being limited to such specifically recited examples and conditions.
[0019] Moreover, all statements herein reciting principles, aspects, and embodiments of the present disclosure, as well as specific examples thereof, are intended to encompass both structural and functional equivalents thereof. Additionally, such equivalents are intended to include both currently known equivalents as well as equivalents developed in the future, i.e., elements developed that perform the same function, regardless of structure.
[0020] Thus, for example, it will be appreciated by those skilled in the art that any block diagrams herein represent conceptual views of illustrative circuitry embodying the principles of the disclosure.
[0021] Unless otherwise specified herein, the figures making up the drawings are not drawn to scale.
[0022] As additional background, we begin by noting that Distributed Fiber Optic Sensing (DFOS) is an important and widely used technique for detecting environmental conditions (temperature, vibration, acoustic excitation vibration, stretch levels, etc.) anywhere along a fiber optic cable that is in turn connected to an interrogator. As is known, a modern interrogator is a system that generates an input signal into the fiber and detects / analyzes the reflected / scattered and subsequently received signal. The signal is analyzed and an output is generated that is indicative of the environmental conditions encountered along the fiber. Such received signals may result from reflections within the fiber such as Raman backscattering, Rayleigh backscattering, Brillion backscattering, etc. DFOS can also use forward signals that exploit the velocity differences of multiple modes. Without loss of generality, the following discussion assumes reflected signals, but the same approach can be applied to forward signals as well.
[0023] FIG. 1(A) is a schematic diagram of a generalized prior art DFOS system. As will be appreciated, a modern DFOS system includes an interrogator that periodically generates optical pulses (or any coded signal) and injects them into an optical fiber. The injected optical pulse signal is transmitted along the optical fiber.
[0024] At locations along the fiber, a small portion of the signal is reflected back to the interrogator. The reflected signal carries information that the interrogator uses to detect, such as, for example, a change in power level indicative of mechanical vibration. Although not shown in detail, the interrogator can include a coded DFOS system that can employ a coherent receiver configuration known in the art, such as that shown in FIG. 1(B).
[0025] The reflected signal is converted to the electrical domain and processed in the interrogator. Based on the time of pulse injection and the time the signal is detected, the interrogator can determine which location on the fiber the signal is coming from and sense the activity at each location on the fiber.
[0026] Those skilled in the art will understand and appreciate that by implementing signal coding on the interrogation signal, more optical power can be transmitted into the fiber, which can advantageously improve the signal-to-noise ratio (SNR) of Rayleigh scattering-based systems (e.g., distributed acoustic sensing, or DAS) and Brillouin scattering-based systems (e.g., Brillouin optical time domain reflectometry, or BOTDR).
[0027] As currently practiced in many modern implementations, a dedicated fiber is allocated to the DFOS system in a fiber optic cable and is physically separated from existing optical communication signals carried on different fibers. However, given the exploding demand for bandwidth, it is becoming increasingly difficult to economically operate and maintain optical fiber solely for DFOS operations. As a result, there is growing interest in integrating communication and sensing systems on a common fiber that is part of a larger multi-fiber cable, or that simultaneously carries live communication traffic in addition to DFOS data.
[0028] Operationally, DFOS systems are assumed to be Rayleigh scattering-based systems (e.g., distributed acoustic sensing, or DAS) and Brillouin scattering-based systems (e.g., Brillouin optical time-domain reflectometry, or BOTDR) that may include coding implementations. In such coding designs, such systems are more likely to be integrated with fiber communication systems due to their low power operation and greater sensitivity to the response time of optical amplifiers.
[0029] In the exemplary arrangement shown in the block diagram, it is assumed that the coded interrogation sequence is generated digitally and modulated onto the sensing laser via a digital-to-analog converter (DAC) and an optical modulator. The modulated interrogation sequence may be amplified to an optimal operating power before being sent to the fiber for interrogation.
[0030] Advantageously, DFOS operations can also be integrated with communication channels via WDM in the same fiber. Within the sensing fiber, the interrogation sequence and the returned sensing signal can be optically amplified by either discrete (EDFA / SOA) or dispersive (Raman) methods. The returned sensing signal undergoes amplification and optical bandpass filtering before being sent to a coherent receiver. The coherent receiver detects the optical fields of both polarizations of the signal and downconverts them to four baseband lanes for analog-to-digital conversion (ADC) sampling and digital signal processor (DSP) processing. As those skilled in the art will readily understand and appreciate, a decoding operation is performed in the DSP to generate the interrogated Rayleigh or Brillouin response of the fiber, and then any changes in that response are identified and interpreted as sensor readings.
[0031] Continuing to refer to the diagram, because the coded interrogation sequence is generated digitally, the out-of-band signal is also generated digitally and then combined with the code sequence before the waveform is created by the DAC. When generated digitally together, the out-of-band signal is only generated outside of the time of the code sequence, so when added together, the combined waveform will have a constant amplitude.
[0032] As will be understood and appreciated by those skilled in the art, DFOS / DAS systems have been shown to detect, record and listen for acoustic vibrations in the audible frequency range, although one of the limiting factors in sensitivity is the physical layout of the fiber optic cables used as sensors.
[0033] For outdoor applications, thick telecommunications-grade fiber optic cables are physically insensitive to low-amplitude vibrations in the audible range, so the quality of the acoustic signal depends heavily on the fiber type, layout, and how the acoustic pressure waves are coupled into the fiber cable.
[0034] As can be readily appreciated, fiber optic cables are widely deployed in both urban and rural areas, advantageously offering greater bandwidth, transmission distances, and reliability.
[0035] To aid in such deployment, a novel underground cable location method based on time series similarity search is disclosed. As will be described, the method of the present invention rapidly searches either specially designed vibration patterns (made with on / off patterns distinguishable from background noise) or actual field vibration signals collected from secondary point sensors (e.g., cell phones, tablets, accelerometers, etc.). Advantageously, it works with different signal patterns generated by different transducers, indicative of different vibration durations, occurring at different locations along the fiber optic cable (e.g., buried or overhead portions of the cable). It provides instant feedback on a resource-limited platform (e.g., laptop or edge device), allowing field technicians to locate underground fiber cables faster and easier than with conventional techniques, without the field technician having to wait more than a few seconds to receive results.
[0036] Limited differences between the present disclosure and the prior art can be highlighted.
[0037] Problems / differences with existing approaches and prior art
[0038] Manual Location and Matching
[0039] A manual search process can be performed by a human visual inspection, but such an operation requires extra effort, is time consuming, and is prone to errors.
[0040] Pattern Recognition by Image Processing
[0041] To automate this process, several image pattern detection methods can be employed. However, traditional edge detection methods only work for patterns with simple shapes and structures, such as vertical bars of a certain width and length on a waterfall image. This pattern can be generated by using a vibrator to continuously vibrate near the fiber for a certain period of time. Depending on the test location, the waterfall image may be very noisy because it contains continuous traffic signals and other vibration signals from the surroundings. To distinguish a simple shape pattern from other sources, it needs to be long and wide enough, which increases the requirements for the test equipment and the length of the test time. Traditional image pattern detectors are difficult to design special structure patterns, cannot obtain consistent results under various conditions, and the complexity negatively affects the computational performance.
[0042] Supervised learning approaches such as neural networks
[0043] Neural network-based image pattern detectors can perform well if the model is pre-trained with a sufficient number of good quality training examples. However, such training is both time-consuming and labor-intensive to collect a sufficiently large dataset in different conditions at different locations, as such signals are not naturally present in waterfall images.
[0044] In contrast to these approaches, our AI module for fast time series template matching automatically detects signal patterns in large-scale waterfall images of entire routes (tens of kilometers). As can be appreciated by those skilled in the art, such operation has the following characteristics: it does not require annotated training data, it works with customized signal patterns, it is robust to variations caused by sensing distance, environmental factors, and the strength of vibration sources, and it provides immediate / real-time results while requiring only limited computing resources.
[0045] As those skilled in the art will understand and appreciate, these features provide many practical advantages.
[0046] Low false alarm rate
[0047] Because our technique does not require training data and can work with any pattern, a set of synthetic signals can be designed in advance and validated with historical daily waterfall data to select a few candidate patterns that yield the lowest false positive rate.
[0048] Customizable patterns
[0049] If the signal later changes, our technique transitions seamlessly without retraining the model.
[0050] Reduced operating time and equipment weight
[0051] The designed signal patterns are more easily detected, even for shorter and weaker patterns or patterns captured from noisy environments. As a result, the overall task can be easily completed by using a portable, small vibration source as the signal source. Field workers will appreciate a lighter, smaller, and more compact vibration source device that is easier to carry than current alternatives. Unfortunately, such vibrators are usually inconsistent and easily affected by other sources, resulting in broken patterns appearing in the waterfall image.
[0052] Real-time feedback The technique and system of the present invention is lightweight, requiring only a mid-level laptop processor. As a result, the necessary calculations are performed locally (no need to send data to a cloud computing system), making real-time results available in the field in real time, a key benefit for the user. As a result, the user can instantly know whether the target cable is underground or not.
[0053] 2 is a schematic diagram outlining a method according to an embodiment of the present disclosure. With reference to this diagram, it can be seen that the hardware elements include:
[0054] Fiber optics, which can use existing installed telecommunications fiber cables or newly installed dedicated fiber.
[0055] Distributed Fiber Optic Sensing System (DFOS), which can be a DAS or DVS, capable of detecting vibration signals along an optical fiber and generating waterfall signal images continuously in real time.
[0056] A handheld vibrating device, such as a hammer drill or vibrator, acts as the signal source.
[0057] A programmable controller that can operate the vibrator in predefined modes or as an audio recorder to collect signals of interest in the field.
[0058] The AI analysis steps include:
[0059] A time series template matching procedure that searches for signal patterns of interest on the raw waterfall data. The time series template matching procedure outputs N regions on the waterfall image, where one of the N regions contains a signal corresponding to the vibration of the test vibrator or none of the N regions contains a signal corresponding to the vibration of the test vibrator. This is a key element of the invention, as the module can operate on any signal pattern without pre-training, and the computations are efficient enough to provide results in real time.
[0060] A region selection procedure that processes the N regions from the previous module. This module selects the regions that contain the signal modified by the probe vibrator. If the probe vibrator does not modify the deformation of the monitoring fiber, no region is selected.
[0061] The operation procedure proceeds as follows:
[0062] Step 1: Monitor the vibration around the fiber cable using the DFOS system.
[0063] For real-time monitoring, the field fiber is connected to a remotely terminated DFOS system.
[0064] Step 2: Pattern design and signal control
[0065] In order to reduce the vibrator's operating time without compromising the visibility of the waterfall image, special patterns can be used. The main idea is to program a metronome that can control the vibration rhythm of the vibrator device, which can generate unique signal patterns on the waterfall image.
[0066] Design pattern pools to maximize detection rates.
[0067] The objective of this step is to generate special patterns with maximum uniqueness at minimum length. When designing the signal, the characteristics of the vibrator (inertia, etc.) and the characteristics of the DFOS must be taken into account. For example, generating a 10Hz square wave is meaningless since it is impossible to control the vibrator to operate at this speed, and if the DFOS pulse rate is low, such a pattern may not be preserved in the waterfall image.
[0068] Note that the amplitude of the patterns is arbitrary and has no impact on the search results, as we advantageously use Pearson's correlation coefficient as a metric with a built-in normalization step.
[0069] FIG. 3 is a graph of an exemplary periodic square wave signal with a 4 second period having on / off cycles of equal length with a sampling rate of 100 Hz, in accordance with an embodiment of the present invention.
[0070] FIG. 4 is a graph of an exemplary periodic square wave signal with a sampling rate of 100 Hz and a 4 second period with on / off cycles of equal length, in accordance with an embodiment of the present disclosure.
[0071] Ambient data is used to select patterns to minimize false alarm rates.
[0072] Historical real-world data is used as test data to evaluate the uniqueness of each design pattern. Uniqueness is defined as the number of subsequences that are highly correlated with the pattern. The test data needs to be as complete as possible to cover a variety of ambient signals from different kinds of background sources.
[0073] Step 3: Use the vibrator as a signal source to generate the designed vibration pattern.
[0074] In this step, the vibrator device is programmed based on the designed pattern so that the signal captured by the DFOS when the vibrator operates is highly correlated with the designed pattern. This can be done by using another hardware, such as a metronome, to control the period of the device's on / off cycles.
[0075] 5(A) and 5(B) are a pair of graphs of amplitude versus time of an actual vibration signal with a specified pattern overlaid after processing with Z-normalization according to an embodiment of the present disclosure. Fig. 5(A) is the ideal case where the vibration signal and the pattern are 94% correlated. Fig. 5(B) is the average case where the signal is corrupted by traffic noise and is only 41% correlated with the pattern.
[0076] Step 4: Template Matching
[0077] A high-level overview of each element is shown in Figure 5(A) and Figure 5(B). As shown in these figures, the template matching module is a key element of the software of the present invention. It processes the waterfall image, which is a 2D matrix with the designed pattern, and can find all signals that correlate with the pattern template.
[0078] We next provide a detailed description of how template matching can be used with distributed optical fiber sensing data.
[0079] Waterfall Data Explained
[0080] As one skilled in the art can appreciate, in an exemplary waterfall image generated from an operational DFOS system, the horizontal axis represents different sensing locations (collected sensing points) on the fiber, the vertical axis represents different timestamps, and the origin is the most recent instant the image was generated. Each pixel (x,y) on the waterfall image is the amplitude of the signal at sensing point x at time y. Each column of pixels can also be treated as a sample of a time series sequence. If there are a total of N columns in the waterfall image, there will be N time series, and template matching is applied to these N time series.
[0081] Time series similarity search
[0082] In this task, we use a similarity search algorithm based on Pearson correlation for template matching. Using the Pearson correlation coefficient as a metric has several advantages:
[0083] First, it is based on lock-step search rather than elastic search like DTW, which is better suited for this task that requires one-to-one coordination as shown in Figure 4(A) and Figure 4(B).
[0084] Second, unlike Euclidean distance, which takes absolute values, correlation only considers the relative trends between two series, which is very important because the amplitude of the signal generated by the same vibrator will be different in different locations in different geological environments underground.
[0085] Finally, from a computational point of view, the algorithm is highly efficient and capable of real-time processing.
[0086] Search Process
[0087] Next, how the search process is performed will be described.
[0088] Assuming we need a query to search, denoted as "q" and assuming it has length m, we need a longer time search "ts" of length n as the search space in which the algorithm will search for the query. The search process is performed using a sliding window held in ts, and since we use Pearson correlation as a metric, the length of this sliding window needs to be the same as q. This sliding window starts initially at index 0 and covers elements of ts from index 0 to index m-1. At each step i, we exclude the first element of the previous window and cover the element at index m-1+i. Figure 6 shows how the sliding window transitions.
[0089] FIG. 6 illustrates an example sliding window that iterates through all n-m+1 sliding windows in ts to find the sliding window position that provides the maximum correlation to the query, according to an embodiment of the present disclosure.
[0090] For each sliding window, we calculate the correlation between the elements in the sliding window and the query. To find the position of the sliding window that maximizes the correlation with the query, we iterate through all n-m+1 sliding windows in ts and consider this subsequence of the sliding window to be the time series most similar to the query in ts.
[0091] 7(A) and 7(B) are a pair of graphs of amplitude versus time of an actual vibration signal from a waterfall image in accordance with an embodiment of the present disclosure, where the vertical lines in FIG. 7(A) are the start and end positions of the sliding window in which the subsequence is most similar to the query, and FIG. 7(B) shows both the query and an overlay of the subsequence.
[0092] We then perform the same operation for each column of the waterfall image. The positions of the most similar subsequences for each column and their Pearson correlation values are stored for further processing in the following steps.
[0093] Performance optimization
[0094] A waterfall image of a 40-second time frame of a 100 km optical fiber cable contains 25 million pixels, so time efficiency needs to be optimized to achieve real-time processing. The state-of-the-art algorithm for general time series similarity search is MASS. MASS is
number
number
[0095] Structurally, our procedure incorporates an automated procedure that estimates the optimal batch size based on the input data size and the hardware platform. The optimized implementation for underground optical fiber localization applications is significantly faster. This allows real-time processing locally on a regular laptop. The performance comparison between the customized implementation and the open source implementation of MASS is the best.
[0096] In summary, the procedure of the present invention employed in the AI model finds the most similar subsequence to the template for each column of the waterfall image. If the waterfall image has N columns, then N tuples are generated, each of which holds the following information: column index, start and end index of the subsequence that is most correlated to the pattern, and the correlation value of said subsequence with the pattern.
[0097] Step 4: Region Estimation
[0098] Based on the results of the previous step, we want to find clusters of sequences that are spatially contiguous and within the same time window, where all sequences in the cluster are highly correlated. For example, such clusters can be represented as regions surrounded by white lines on the waterfall image. When estimating the vibration signal of interest in the next step, regions are of final interest instead of individual subsequences, since they contain more information than individual sequences. Such regions exist because disturbances affect a range of fiber cables that contain multiple sensing points.
[0099] To estimate such regions, we use the results of the previous step. We iterate through the tuples in descending order based on the correlation value, and the iteration stops when the required number of regions is reached. For each tuple, the column index, row index, and Pearson correlation value of the subsequence are recorded. For each tuple, we follow the flow diagram in Figure 8.
[0100] FIG. 8 illustrates an example flow diagram illustrating tuple iteration according to an aspect of the present disclosure.
[0101] To make the localized region more complete, we use the actual sequence as a template instead of the synthetic design pattern to search its neighboring space. Although the synthetic design pattern in ideal situations (high SNR and low noise) also works fine, in most cases the signal is distorted by other background noises such as traffic.
[0102] Step 5: Area selection
[0103] In this final step, a region corresponding to the vibration signal of interest is selected, or none is selected if none of the regions from the previous steps correspond to a vibration signal.
[0104] We describe two approaches for this step: a threshold-based method and a region overlap method.
[0105] Threshold-based approach
[0106] In the threshold-based method, the following values are first calculated for each region and then determined based on a threshold: SNR, regional Pearson correlation, and autocorrelation.
[0107] For SNR, the signal-to-noise ratio of this region and the surrounding regions. The next value calculated is the similarity of the region to the designed pattern, which is calculated as the Pearson correlation value of the spatially and temporally smoothed subsequence of this region to the designed pattern. The spatially smoothed sequence is generated by adding (vector-wise) the entire time series from each column in the region and dividing by the number of columns. Then, a sliding average window is used to perform temporal smoothing on the spatially smoothed sequence.
[0108] Similar to the calculation above, we calculate both the top-correlated sequence and a spatiotemporal smoothed version. Here, the full time series is used instead of a subsequence. This allows us to consider a larger time range, since the signal is usually generated by a stationary source of oscillations from the surrounding area and is usually long-lasting. The result is compared to the autocorrelation of the design pattern filled with zeros, and the similarity is still measured by Pearson's correlation.
[0109] Finally, a threshold is set for each measurement to evaluate whether the region is a region of interest, in other words whether the signal in this region is caused by the probe vibrator.
[0110] Overlapping Areas
[0111] Another approach is region overlap, which is done by searching two or more designed patterns separately at the same location. The idea behind this is that the designed patterns should appear at the same location on the waterfall image. Also, it is highly unlikely that one background noise region is correlated with all the designed patterns, i.e., the top few estimated regions of different design patterns should be separated except for the region of interest.
[0112] FIG. 9 is an exemplary flow diagram illustrating the overall operation of systems and methods according to aspects of the present disclosure.
[0113] With reference to this diagram, note that the procedure of the present invention begins by connecting the DFOS system from the control (central) station to the field test fiber cable.
[0114] Next, an engineer will carry out an on-site survey with a vibration generator.
[0115] Such a survey involves a technician activating a vibration generator near the test fiber cable path, for example in a manhole where the field test fiber is deployed. This operation involves generating / transmitting a named sequence pattern of unique on / off (vibration) patterns / signatures (e.g., on for 5 seconds, off for 5 seconds, etc.) from a central office over Wifi / LTE / 5G / etc. to the field vibration generator.
[0116] The vibration patterns, together with ambient noise such as road traffic and other vibration sources occurring along the test fiber in the field, result in characteristic scattering that is detected during DFOS operation.
[0117] The operation of the DFOS and associated analysis system is provided by the central station and involves recognizing the unique vibration patterns mechanically influenced by vibration sources in the field and detected by the test fiber in the field. AI algorithms in machine learning procedures are then able to identify the distances associated with the vibrations generated in the field.
[0118] These distances are correlated with the GPS coordinates of the cable distance data to generate, among other items, a graphical location display / data.
[0119] The graphical locations so determined may then be displayed on a graphical user interface to direct a subsequent technician to such locations.
[0120] While the present disclosure has been presented using some specific examples, those skilled in the art will recognize that the present teachings are not so limited. Accordingly, the present disclosure should be limited only by the scope of the appended claims.
Claims
1. A method for underground cable location using fast time series template matching and distributed fiber optic sensing (DFOS), comprising: An optical sensor fiber; a DFOS interrogator in optical communication with the optical sensor fiber, the DFOS interrogator configured to generate optical pulses, couple the generated pulses into the optical sensor fiber, and receive backscattered signals from the optical sensor fiber; an intelligent analyzer configured to analyze the DFOS data received by the DFOS interrogator and determine from the backscattered signals vibrational activity occurring at a location along the optical sensor fiber; providing a DFOS system including: disposing a programmable vibration generator at a field location proximate to said optical sensor fiber; transmitting to the programmable vibration generator a unique vibration pattern to be generated by the vibration generator; activating the programmable vibration generator to generate the transmitted unique vibration pattern; operating the DFOS system and collecting / analyzing the determined vibration activity to further determine vibration activity indicative of the unique vibration pattern produced by the vibration generator.
2. The method of claim 1 , further comprising associating the vibration activity indicative of the unique vibration pattern generated by the vibration generator with a physical location along the optical sensor fiber.
3. The method of claim 2 , wherein the physical location is correlated to a GPS coordinate.
4. 4. The method of claim 3, further comprising repositioning the vibration generator to a different location along the optical sensor fiber and transmitting a different vibration pattern to the vibration generator to generate, and activating the vibration generator and a DFOS system to determine the new location of the vibration generator.
5. The method of claim 3 , further comprising generating the unique vibration pattern that is transmitted to the vibration generator.
6. The method of claim 3 , further comprising generating waterfall data from the backscattered signals.
7. The method of claim 6 , wherein determining vibration activity indicative of the unique vibration pattern produced by the vibration generator comprises pattern matching of the waterfall image.
8. The method of claim 7 , wherein the pattern matching comprises correlating a 2D matrix having the designed pattern with a pattern template.
9. The method of claim 8 , wherein the waterfall data image includes N columns of N time series, and the pattern matching is applied to all of the N time series.
10. The method of claim 9 , wherein the pattern matching comprises determining a similar subsequence of the pattern template for each of the N columns of the waterfall data image.
Citation Information
Patent Citations
Sensing device for preventing damage of underground structure
JP2001059719A
Manhole position specifying method and manhole position specifying system
JP2020052030A
Smart optical cable positioning / location using optical fiber sensing
US20200124735A1
Systems and methods for determining fiber cable geographic locations
US20210278314A1
Sparse excitation method for 3-dimensional underground cable localization by fiber optic sensing
US20210311186A1
