Optical fiber fault locating method and device based on OTDR reflection image

Abstract

This application relates to the field of optical fiber fault detection technology, and discloses a method and apparatus for optical fiber fault location based on OTDR reflection images. The method includes: controlling an OTDR device to send test light pulses to the optical fiber link under test to obtain reflected signals and generate a reflection map; identifying strong reflection events from the reflection map and determining the corresponding obscured areas; establishing a strong reflection tail signal model for the obscured areas, and compensating and reconstructing the reflected signals within the obscured areas based on the strong reflection tail signal model to generate a compensated reflection map; extracting feature vectors of the reflection mapping from the reflection map and the compensated reflection map respectively and fusing them; determining the fault probability value of candidate locations based on the fused feature vectors, and thus determining the fault point in the optical fiber link. This invention, by modeling the influence of strong reflection tails and compensating and reconstructing the signals in the obscured areas, can recover weak fault features obscured by strong reflection tails, improving the accuracy and efficiency of optical fiber fault location.

Description

Technical Field

[0001] This application relates to the field of optical fiber fault detection technology, specifically to an optical fiber fault location method and apparatus based on OTDR reflection images. Background Technology

[0002] An optical time domain reflectometer (OTDR) is an important instrument used to detect the quality of fiber optic links and locate fiber optic faults. An OTDR generates a reflection map (OTDR trace) by sending test optical pulses into the fiber optic link and measuring the Rayleigh scattering signal generated during the propagation of the pulses in the fiber, as well as the Fresnel reflection signals generated at connectors, fusion splices, or fiber breaks. By analyzing the peaks, valleys, and attenuation trends in the reflection map, connectors, fusion splices, and possible fault locations in the fiber optic link can be identified.

[0003] However, in actual fiber optic links, strong Fresnel reflections are often generated by components such as connectors, adapters, splitter inputs, and flanges, resulting in strong reflection peaks in OTDR reflection maps. Due to the limited dynamic range of OTDR devices, when a strong reflection peak appears, the reflected signal behind it may be affected by tailing, ringing, or energy diffusion, thus masking subsequent weak reflection characteristics. This is especially true when there are weak faults in the fiber optic link, such as micro-bending loss, minor cracks, or localized compression. These faults are inherently weak in the reflection map, and if they occur after a strong reflection event, they are even more difficult to identify.

[0004] Therefore, accurately identifying the weak fault characteristics behind a strong reflection event is a key technical problem that needs to be solved to improve the efficiency and accuracy of OTDR fault location. Summary of the Invention

[0005] The purpose of this application is to provide a method, apparatus, and computer device for fiber optic fault location based on OTDR reflection images, so as to solve the problems mentioned in the background art.

[0006] According to a first aspect of this application, a fiber optic fault location method based on OTDR reflection images is provided, comprising: The OTDR device is controlled to send test light pulses to the fiber optic link under test to obtain a series of reflection signals, and a reflection pattern is generated based on the series of reflection signals; Strong reflection events are identified from the reflection map, and corresponding occlusion intervals are determined based on the strong reflection events; A strong reflection trailing signal model is established for the occluded area. Based on the strong reflection trailing signal model, the reflection signal of the reflection map in the occluded area is compensated and reconstructed to generate a compensated reflection map. Reflection mapping feature vectors are extracted from the reflection map and the compensated reflection map respectively and fused. The fault probability value of the candidate location is determined based on the fused feature vector, and the fault point in the optical fiber link is determined according to the fault probability value.

[0007] According to a second aspect of this application, an optical fiber fault location device based on OTDR reflection images is provided, the device comprising: The reflection pattern generation module is used to control the OTDR device to send test light pulses to the fiber optic link under test to obtain a series of reflection signals, and to generate a reflection pattern based on the series of reflection signals; The occlusion interval determination module is used to identify strong reflection events from the reflection map and determine the corresponding occlusion interval based on the strong reflection events; The compensation and reconstruction module is used to establish a strong reflection trailing signal model for the occluded area, and to compensate and reconstruct the reflection signal of the reflection map in the occluded area based on the strong reflection trailing signal model to generate a compensated reflection map. The fault location module is used to extract feature vectors of reflection mapping from the reflection map and the compensated reflection map respectively and fuse them, determine the fault probability value of the candidate location based on the fused feature vector, and determine the fault point in the optical fiber link according to the fault probability value.

[0008] According to a third aspect of this application, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program that, when executed by the processor, implements the method as described in any of the preceding claims.

[0009] Compared with existing technologies, the fiber optic fault location method based on OTDR reflection images provided by this invention identifies strong reflection events and determines the corresponding obstruction intervals in the reflection image, establishes a strong reflection tail signal model, compensates and reconstructs the reflection signals within the obstruction interval, thereby generating a compensated reflection image. Based on the original reflection image and the compensated reflection image, reflection mapping feature vectors are extracted and fused to calculate the fault probability value of candidate locations and determine the fault point. By modeling and compensating for the impact of strong reflection tails, the obstruction effect of strong reflection events on subsequent signals can be effectively reduced, hidden weak fault characteristics can be restored, and the ability to identify weak faults such as microbending loss and connector damage can be improved. This enhances the accuracy and detection efficiency of fiber optic fault location, reduces the need for repeated scanning and manual comparison of the entire fiber optic link, and improves the reliability and efficiency of fiber optic network operation and maintenance. Attached Figure Description

[0010] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0011] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 A flowchart illustrating a fiber optic fault location method based on OTDR reflection images provided in an embodiment of this application; Figure 2 A schematic diagram of the structure of an optical fiber fault location device based on OTDR reflection images provided in this application embodiment; Figure 3 This is a schematic diagram of the structure of the shading area determination module 20 provided in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of the compensation and reconstruction module 30 provided in the embodiments of this application; Figure 5 This is a schematic diagram of the fault location module 40 provided in an embodiment of this application. Detailed Implementation

[0012] For ease of understanding, the following explanation will focus on the fiber optic link being monitored in a fiber optic communication network.

[0013] The solution in this embodiment can be deployed on the server corresponding to the OTDR monitoring platform, the cloud diagnostic platform, or the intelligent fiber optic distribution scheduling and management system. It is used to analyze the reflection map of the fiber optic link, identify faults, and output location results. The system can interface with OTDR devices, link resource management platforms, and operation and maintenance alarm platforms, providing a data acquisition, compensation calculation, fault display, and maintenance dispatching entry point for this invention. It should be noted that the fiber optic link in the following embodiments can be a trunk optical cable, access fiber, data center interconnect fiber, or a complex link containing multiple connectors / splitters / distribution units. This invention is not limited by the physical form of the link, but focuses on the recovery and location of weak fault characteristics under strong reflection and obstruction phenomena in the OTDR reflection map.

[0014] like Figure 1 As shown in the figure, this application discloses a fiber optic fault location method based on OTDR reflection images, including: S1, control the OTDR device to send test light pulses to the fiber optic link under test to obtain a series of reflection signals, and generate a reflection pattern based on the series of reflection signals; In this step, a test light pulse is emitted into the fiber optic link under test using an OTDR device, and the scattered / reflected signals returned at each sampling time are collected to form a reflection map unfolded along the distance axis. It can be understood that after emitting the pulse, the OTDR device records the echo return time and the corresponding echo power value, and calculates the corresponding distance position of each sampling point in the link based on the propagation time and the fiber refractive index, thus forming a two-dimensional distance-reflection power relationship curve. This curve can be represented by a discrete sampling point sequence, or it can be further plotted by the monitoring platform as a reflection image for subsequent image analysis and feature extraction.

[0015] OTDR devices can perform measurements according to preset pulse width, sampling resolution, averaging times, and measurement dynamic range. For example, a narrower pulse width can be used to improve spatial resolution for shorter links, while a wider pulse width can be used to improve measurement dynamic range for longer links; however, regardless of the measurement parameters used, the present invention can identify and compensate for areas with strong reflections on the reflective layer surface. For example, let the round-trip time corresponding to the kth sampling point be... If the refractive index of the optical fiber is n, then its distance from the location is... can be This is determined, where c is the speed of light. From this, the reflected signal sequence can be obtained. ,in This represents the reflected power value at the corresponding distance. The monitoring platform can further plot this reflected signal sequence as a reflection map in distance order.

[0016] To improve the stability of subsequent strong reflection event identification and compensation reconstruction, basic preprocessing of the original reflection signal can be performed before generating the reflection map. This includes interpolating for abnormally missing points, removing isolated noise points that significantly exceed the physical range, slightly smoothing the power curve, and aligning the distance axis of reflection signals from different measurements. Understandably, this alignment can be achieved by referencing the link start point, the first significant event point, or the known connector location.

[0017] For example, when performing two consecutive measurements on the same connector on the same link, if the strong reflection peak is measured at 1002.4m in the first measurement and 1002.8m in the second measurement, the two measurement results can be finely adjusted and aligned based on the fixed starting point of the link and the known refractive index, so that the two reflection patterns are consistent near this position, so that subsequent calculations of parameters such as peak width, trailing, and shielding area length can be performed more stably.

[0018] The above processing can reduce the impact of sampling jitter, instantaneous equipment noise, or differences in repeated measurements on the consistency of subsequent judgments.

[0019] S2, identify strong reflection events from the reflection map, and determine the corresponding occlusion interval based on the strong reflection events; Understandably, in actual fiber optic links, locations such as connector end-face contamination, flange interfaces, adapter misalignment, movable connection points, and some splitter inputs typically create sharp reflection peaks in OTDR reflection maps with amplitudes significantly higher than the background. When the amplitude of these reflection peaks is high, the signals behind them do not immediately fall back to a stable background; instead, they exhibit a certain length of trailing, energy diffusion, or ringing effects, thus masking subsequent weak fault characteristics. This invention first identifies such strong reflection events and then determines the obscured area based on the range of their trailing effect, allowing for targeted compensation in subsequent steps.

[0020] In some embodiments, identifying strong reflection events from the reflection map includes: S21, perform local peak detection on the reflection map to obtain candidate peak points, and calculate the peak height and peak width of each candidate peak point; wherein, the peak height is the difference between the power value of the candidate peak point and the background power value of its neighborhood, and the peak width is the difference between the corresponding distance positions when the power on the left and right sides of the candidate peak point drops to a preset proportion of the peak power; Specifically, local peak detection can be performed using a sliding window scanning method. For example, a window of length w slides point by point along the distance axis. When the power value of the sampling point at the center of the window is simultaneously greater than the power values ​​of several sampling points in its left and right neighborhoods, and its power value is higher than the local background level, this point can be considered a candidate peak point. The neighborhood background power value can be determined using the mean, median, or smoothed background baseline value of the non-peak segments within a certain distance to the left and right of the candidate peak point.

[0021] Peak height can be expressed as ,in The power value at the candidate peak point. This represents the background power value of the neighborhood. The peak width can be determined based on the left and right distance differences corresponding to half the peak height, one-third of the peak height, or other preset proportions. For example, the distance difference between the left and right intersection points when the power drops to 50% of the peak power can be used as the peak width.

[0022] S22, when the peak height is greater than the first threshold and the peak width is less than the second threshold, the corresponding candidate peak point is determined as a strong reflection event.

[0023] Strong reflection events typically exhibit both high peak values ​​and relatively concentrated localized spikes. If only peak height is considered, some broad, gentle baseline fluctuations or long, gentle slopes may be misidentified; similarly, if only peak width is considered, localized sharp noise may also be misidentified. Therefore, this embodiment uses both peak height and peak width.

[0024] Specifically, by requiring peak height to be greater than a first threshold and peak width to be less than a second threshold, it is possible to more effectively distinguish between genuine strong reflection events and ordinary power fluctuations. For example, if a candidate peak point is 9 dB higher than the background and its peak width is 2 m, it can be determined that it meets the conditions for a strong reflection event; if another candidate peak point is only 3 dB higher, even if the peak width is small, it does not constitute a strong reflection event sufficient to form obvious shading; if a peak is 8 dB higher but its peak width reaches 20 m, it is more likely to be a slow power drift or segment transition, rather than a local strong reflection.

[0025] In some embodiments, determining the corresponding occlusion zone based on the strong reflection event includes: S23, obtain the peak position and peak height corresponding to the strong reflection event; calculate the trailing attenuation rate within a preset observation distance range behind the peak position; Specifically, for each identified strong reflection event, the peak position and corresponding peak height of that event in the reflection map are obtained. The peak position can be represented as the distance coordinates corresponding to the top of the strong reflection peak, for example, denoted as... The peak height is the difference between the power value at the peak point and the background power value in its neighborhood, denoted as . .

[0026] After obtaining the peak position, a preset observation distance range is selected behind the peak position to analyze the attenuation trend of the trailing signal. The preset observation distance range can be set based on the fiber optic link length, OTDR measurement resolution, or empirical statistical results; for example, a range of 10m to 50m behind the peak position can be used as the observation interval. Within this observation interval, the trailing signal typically exhibits a gradual attenuation trend.

[0027] Subsequently, the variation in reflected power within the observation interval is calculated to obtain the tail attenuation rate. The tail attenuation rate characterizes the speed at which the power curve recovers to the background level after a strong reflection event. For example, two distance locations within the observation interval can be selected. and Obtain their corresponding reflection power. and The trailing attenuation rate is calculated as follows:

[0028] in, Indicates the trailing decay rate. and All are located within the observation interval after the peak position.

[0029] To improve computational stability, preferably, multiple sampling points can be selected within the observation interval, and the overall attenuation trend can be estimated using linear or exponential fitting. For example, least-squares fitting can be performed on the reflection power curve within the observation interval to obtain the average attenuation rate. This method reduces the impact of local noise or minor anomalies on the attenuation rate calculation results.

[0030] S24, calculate the shading length based on the peak height and the trailing attenuation rate, and determine the interval from the peak position to the peak position plus the shading length as the shading interval; wherein, the shading length is positively correlated with the peak height and negatively correlated with the trailing attenuation rate.

[0031] The higher the peak value of strong reflection, the stronger its trailing signal is usually, and the more significant the shading effect on subsequent reflected signals. Conversely, a higher trailing attenuation rate indicates that the trailing signal attenuates faster, and its influence range is relatively shorter. Therefore, the shading length should be positively correlated with the peak height and negatively correlated with the trailing attenuation rate.

[0032] The shielding length can be calculated based on the peak height and the trailing attenuation rate. For example, it can be calculated in the following form:

[0033] in, Indicates the length of the shielding. Indicates the peak height of a strong reflection event. Indicates the trailing decay rate. This is an empirical coefficient used to adjust based on the characteristics of different OTDR devices or link conditions.

[0034] Using the above calculation method, when the peak height is large or the trailing attenuation rate is small, the calculated shading length will increase accordingly; conversely, when the peak height is small or the trailing attenuation rate is large, the shading length will decrease accordingly.

[0035] Once the shielding length is obtained, the corresponding shielding interval can be determined. Specifically, the location of the strong reflection peak can be determined. As the starting point of the shielding interval, and As the endpoint of the occlusion interval, the occlusion interval is obtained: .

[0036] The shielding range refers to the distance within which a strong reflection event significantly affects the reflected signal behind it. Within this range, the signal in the original reflection map may be affected by the superposition of trailing signals. Therefore, in subsequent steps, it is necessary to compensate and reconstruct the reflected signal within this range to restore the true link reflection characteristics.

[0037] To ensure consistency in the determination of the occlusion interval across different measurements, it is preferable to set reasonable upper and lower limits for the occlusion length. For example, when the calculated occlusion length is too large, a maximum occlusion length threshold can be set to avoid overexpansion; when the calculated occlusion length is too small, a minimum occlusion length can be set to ensure that subsequent compensation processing still has sufficient range.

[0038] S3, establish a strong reflection trailing signal model for the shading area, and compensate and reconstruct the reflection signal of the reflection map in the shading area based on the strong reflection trailing signal model to generate a compensated reflection map. Because strong reflection events produce residual system responses in OTDR systems, these residual signals typically appear as a trailing signal that gradually decays behind the strong reflection peak. When this trailing signal is superimposed on the actual link reflection signal, it masks weak reflection characteristics within the shielded area, making it difficult to identify weak faults such as micro-bending loss, minor connection damage, or localized abnormal bending in the original reflection map. Therefore, this step establishes a strong reflection trailing signal model to estimate its impact, and then subtracts the estimated trailing signal as a compensation term from the original reflection signal, thereby restoring a reflection response within the shielded area that more closely resembles the actual link structure.

[0039] In some embodiments, a strong reflection trailing signal model is established for the shaded area, including: S31, The strong reflection trailing signal is modeled using an exponential decay function, so that the strong reflection trailing signal is related to the peak height and the attenuation coefficient; wherein, the attenuation coefficient is obtained by performing least squares fitting on the reflection signal in the shading interval; Specifically, for the shaded area The power curve after the strong internal reflection peak is modeled. Since the OTDR system response typically exhibits exponential decay characteristics after strong reflection, an exponential decay function can be used to describe the tail signal. The tail signal model is expressed as:

[0040] in: Indicates distance position The estimated value of the trailing signal at the location; Indicates the peak height of a strong reflection event; Indicates the location of the peak reflection; This represents the trailing attenuation coefficient.

[0041] In this model, the amplitude of the trailing signal is related to the peak height of the strong reflection. The attenuation rate of the trailing signal is proportional to the attenuation coefficient. The larger the attenuation coefficient, the faster the trailing signal attenuates and the shorter its shadowing range; the smaller the attenuation coefficient, the longer the trailing effect lasts.

[0042] To obtain a reasonable attenuation coefficient, this embodiment uses a least-squares fitting method to fit the reflected signal within the shielded region. Specifically, multiple sampling points within the shielded region can be selected. The attenuation coefficient is determined by minimizing the following error function:

[0043] in, Indicates the original reflectance at a distance position The reflected power value at that location.

[0044] To improve fitting stability, a relatively smooth region following the peak position can be preferentially selected for fitting, and the weight of obviously abrupt abrupt changes can be reduced. For example, if there is a significant local power drop or rise within the shading interval, this position may correspond to a real link event and should not be fully included in the fitting range of the tailing model. By reasonably selecting the fitting sample points, the tailing model can more accurately represent the residual system response caused by strong reflection events.

[0045] Correspondingly, based on the strong reflection trailing signal model, the reflection signal of the reflection map within the shielded area is compensated and reconstructed to generate a compensated reflection map, including: S32, the tail estimation signal is calculated in the shading area using the strong reflection tail signal model, and the tail estimation signal is used as a compensation term to correct the reflection signal in the shading area to obtain a compensation signal.

[0046] Specifically, after obtaining the trailing signal model, the estimated trailing signal can be calculated point-by-point within the occluded interval. For any sampling point within the occluded interval... Its tail estimation signal can be expressed as:

[0047] Subsequently, the estimated trailing signal is subtracted from the original reflected signal as a compensation term to obtain the compensated reflected signal:

[0048] in: Indicates the original reflected signal; This indicates the tail estimation signal; This represents the reflected signal after compensation.

[0049] By applying the above point-by-point compensation, the impact of strong reflection tails on subsequent signals can be effectively reduced, making the power curve within the shielded section closer to the reflection response generated by the actual link structure. For example, in some cases, weak fault features located about 10m behind the strong reflection peak in the original reflection map may be obscured by the tail background. However, after the above compensation process, the local curvature change or power drop characteristics at this location will be more obvious, thereby improving the visibility of weak fault identification.

[0050] S33, when the compensation signal is lower than the noise baseline, the compensation signal is interpolated and corrected based on adjacent sampling points to generate the compensation reflection map.

[0051] Specifically, after compensation calculations, the compensated signal at some sampling points may be lower than the system noise baseline. This phenomenon is usually due to the tailing model being slightly higher than the actual tailing level at certain locations, resulting in overcorrection during the compensation process.

[0052] To ensure the stability of the compensation results, the compensation signal is further interpolated and corrected. Specifically, when the compensation signal at a certain sampling point is detected to be lower than the noise baseline, interpolation estimation can be performed on that location based on adjacent sampling points. For example, the following linear interpolation method can be used:

[0053] in, and These represent the compensation signals at the adjacent positions before and after the current sampling point.

[0054] It should be noted that spline interpolation, local polynomial fitting, or moving average can also be used to correct the compensation signal, thereby obtaining a smoother power curve that conforms to physical laws. By performing the above interpolation correction process, unreasonable negative values ​​or abrupt changes in the compensation signal can be avoided, ensuring that the compensation result remains numerically continuous and conforms to the actual variation law of the reflected signal in the optical fiber link. Finally, by performing the above compensation and correction process on all sampling points within the shielded area, a compensated reflection map can be generated.

[0055] Compared with the original reflection map, the compensated reflection map can effectively reduce the background influence of strong reflection tails, making the weak fault features in the occluded area more clearly presented, thus providing a more reliable data foundation for the subsequent extraction of reflection mapping feature vectors and calculation of fault probability.

[0056] S4. Extract reflection mapping feature vectors from the reflection map and the compensated reflection map respectively and fuse them. Determine the fault probability value of the candidate location based on the fused feature vector, and determine the fault point in the optical fiber link according to the fault probability value.

[0057] The original reflection map preserves the complete measurement response of the fiber optic link, while the compensated reflection map reduces the obscuring effect of strong reflection tails on subsequent signals, thus making weak fault characteristics more obvious. To fully utilize the information contained in the two reflection maps, this step extracts the feature vectors of the reflection maps from the original and compensated reflection maps respectively, and constructs a comprehensive representation through feature fusion to further calculate the fault probability value of candidate locations, thereby achieving accurate identification of fiber optic fault locations.

[0058] In some embodiments, extracting and fusing reflection mapping feature vectors from the reflection map and the compensated reflection map respectively includes: S41, extract at least three features from the peak height, peak width, local attenuation slope, second curvature and local energy of the reflection map along the distance axis using a sliding window to form a first feature vector, and extract the corresponding features from the compensated reflection map to form a second feature vector. Specifically, a length of [value] is set in the distance axis direction. A sliding window is used, moving point by point along the reflection map to extract features of the reflected signal within each window region. In this way, the local reflection characteristics of the entire fiber optic link can be continuously analyzed.

[0059] For example, the following features can be extracted from each window region: 1) Peak height: Used to characterize the degree to which a local reflection peak protrudes relative to the background signal. For example, it can be expressed as:

[0060] in, This represents the maximum power value within the window. This represents the average background power within the window.

[0061] 2) Peak width: Used to characterize the extent to which a local reflection peak extends along the distance axis. For example, it can be calculated by the difference in distance between the left and right sides when the power drops to a certain percentage (e.g., 50%) of the peak value.

[0062] 3) Local attenuation slope: Used to reflect the trend of power variation with distance within the window area, and can be expressed as:

[0063] in, and This represents the distance between the two ends of the window.

[0064] 4) Second-order curvature: Used to describe the degree of curvature of a curve, it can be expressed as: When this value is large, it usually means that there is a significant local variation at that location, such as microbending loss or connector damage.

[0065] 5) Local energy: Used to reflect the overall signal strength within the window, for example, it can be represented as:

[0066] in, This indicates the location of each sampling point within the window.

[0067] By extracting at least three of the above features, a first feature vector can be constructed: .

[0068] Similarly, the same sliding window method is used to extract the corresponding features on the compensated reflection map to form the second feature vector:

[0069] Understandably, the feature vectors extracted from the original reflection map better reflect the overall structure of the real link, while the feature vectors extracted from the compensated reflection map better highlight the weak fault features that are masked by strong reflection tails.

[0070] For example, in some links, microbending losses located about 12m behind strong reflection peaks may only appear as slight power changes in the original reflection map, but may appear as significant local curvature changes in the compensated reflection map. By extracting both feature vectors simultaneously, the ability to identify this type of weak fault can be improved.

[0071] S42, the first feature vector and the second feature vector are concatenated or weighted and fused to obtain a fused feature vector; Specifically, feature splicing can be used for fusion. This means directly combining the two feature vectors. Alternatively, a weighted fusion method can also be used. ;in, For fusion weighting coefficients.

[0072] By using the above method, information from both the original reflection map and the compensated reflection map can be used simultaneously, so that the fused feature vector can reflect both the overall structure of the link and highlight the characteristics of weak faults.

[0073] For example, for some normal link regions, the feature values ​​in the original reflection map and the compensated reflection map usually have high consistency; however, for locations with weak faults, the feature changes in the compensated reflection map will be more obvious, thus making the fused feature vector show a more significant difference at that location.

[0074] S43, the fused feature vector is input into the machine learning classification model to output the fault probability value of the candidate position, and when the fault probability value exceeds the preset probability threshold, the corresponding candidate position is determined as the fault point.

[0075] Specifically, the machine learning classification model can be a logistic regression model, a support vector machine model, a random forest model, or a neural network model, etc. This invention does not limit the type of model.

[0076] For example, in a logistic regression model, the failure probability value can be expressed as:

[0077] in, To fuse feature vectors, For the model weight vector, This is a bias term.

[0078] When the fault probability value output by the model is greater than the preset probability threshold (e.g., 0.7 or 0.8), the corresponding candidate location can be identified as a fault point.

[0079] To improve detection stability, cluster analysis can be performed on multiple adjacent high-probability locations. For example, when the failure probability of multiple window center points within a certain interval exceeds a threshold, the interval can be merged into a single fault region, and the center location of this region can be taken as the final fault point. For instance, when the failure probabilities of multiple window center points within the range of 1002m to 1004m are 0.81, 0.86, and 0.83 respectively, the interval can be merged into a single fault event, and 1003m can be output as the fault location.

[0080] By following the steps above, the fault point in the optical fiber link can be determined based on the fused feature vector.

[0081] Please see Figure 2 This application also provides an optical fiber fault location device 100 based on OTDR reflection images, the device comprising: The reflection pattern generation module 10 is used to control the OTDR device to send test light pulses to the fiber optic link under test to obtain a series of reflection signals, and generate a reflection pattern based on the series of reflection signals; The occlusion interval determination module 20 is used to identify strong reflection events from the reflection map and determine the corresponding occlusion interval based on the strong reflection events; The compensation and reconstruction module 30 is used to establish a strong reflection trailing signal model for the shading area, and to compensate and reconstruct the reflection signal of the reflection map in the shading area based on the strong reflection trailing signal model to generate a compensated reflection map. The fault location module 40 is used to extract feature vectors of reflection mapping from the reflection map and the compensated reflection map respectively and fuse them, determine the fault probability value of the candidate location based on the fused feature vector, and determine the fault point in the optical fiber link according to the fault probability value.

[0082] In some embodiments, please refer to Figure 3The occlusion zone determination module 20 includes a strong reflection recognition unit 201, which is used for: Local peak detection is performed on the reflection map to obtain candidate peak points, and the peak height and peak width of each candidate peak point are calculated respectively; wherein, the peak height is the difference between the power value of the candidate peak point and the background power value of its neighborhood, and the peak width is the difference between the corresponding distance positions when the power on the left and right sides of the candidate peak point drops to a preset proportion of the peak power; When the peak height is greater than the first threshold and the peak width is less than the second threshold, the corresponding candidate peak point is determined as a strong reflection event.

[0083] In some embodiments, please continue to refer to Figure 3 The shading interval determination module 20 further includes a shading interval calculation unit 202, which is used for: Obtain the peak position and peak height corresponding to the strong reflection event, and calculate the trailing attenuation rate within a preset observation distance range behind the peak position; The shading length is calculated based on the peak height and the trailing attenuation rate, and the interval from the peak position to the peak position plus the shading length is determined as the shading interval; wherein, the shading length is positively correlated with the peak height and negatively correlated with the trailing attenuation rate.

[0084] In some embodiments, please refer to Figure 4 The compensation and reconstruction module 30 includes: The trailing modeling unit 301 is used to model the strong reflection trailing signal using an exponential decay function, so that the estimated value of the strong reflection trailing signal is related to the peak height and attenuation coefficient of the strong reflection event; wherein the attenuation coefficient is obtained by performing least squares fitting on the reflection signal within the shading interval. The compensation and correction unit 302 is used to calculate the estimated trailing signal in the shading area using the strong reflection trailing signal model, and to use the estimated trailing signal as a compensation term to correct the reflection signal in the shading area to obtain a compensation signal. The interpolation correction unit 303 is used to interpolate and correct the compensation signal based on adjacent sampling points when the compensation signal is lower than the noise baseline, so as to generate the compensation reflection map.

[0085] In some embodiments, please refer to Figure 5 The fault location module 40 includes: The feature extraction unit 401 is used to extract at least three features from the peak height, peak width, local attenuation slope, second curvature and local energy of the reflection map along the distance axis using a sliding window to form a first feature vector, and to extract the corresponding features from the compensated reflection map to form a second feature vector. The feature fusion unit 402 is used to concatenate or weightedly fuse the first feature vector and the second feature vector to obtain a fused feature vector; The probability output and localization unit 403 is used to input the fused feature vector into the machine learning classification model to output the fault probability value of the candidate position, and to determine the corresponding candidate position as the fault point when the fault probability value exceeds a preset probability threshold.

[0086] This application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements the method as described in any of the preceding claims.

[0087] The computer device can be a server, industrial control computer, network monitoring terminal, cloud computing node, or other computing device with data processing capabilities. The memory can be a non-volatile or volatile storage medium such as random access memory (RAM), read-only memory (ROM), flash memory, hard disk, or solid-state drive, used to store computer programs and related data. The processor can be a central processing unit (CPU), digital signal processor (DSP), graphics processing unit (GPU), or dedicated integrated processing chip, used to execute the computer program to implement the aforementioned fiber optic fault location method.

[0088] In some embodiments, the computer device may further include a communication interface for data interaction with an OTDR device, a network management system, or a remote operation and maintenance platform, thereby acquiring reflected signal data of the fiber optic link and outputting fault location results. Through the above structure, the computer device can automatically analyze OTDR reflection maps and locate faults when executing the computer program, thereby improving the efficiency of fiber optic network operation and maintenance.

[0089] The above description represents the preferred embodiments of the present invention. It should be noted that, for those skilled in the art, various improvements and modifications can be made without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A fiber optic fault location method based on OTDR reflection images, characterized in that, Includes the following steps: The OTDR device is controlled to send test light pulses to the fiber optic link under test to obtain a series of reflection signals, and a reflection pattern is generated based on the series of reflection signals; Strong reflection events are identified from the reflection map, and corresponding occlusion intervals are determined based on the strong reflection events; A strong reflection trailing signal model is established for the occluded area. Based on the strong reflection trailing signal model, the reflection signal of the reflection map in the occluded area is compensated and reconstructed to generate a compensated reflection map. Feature vectors of the reflection mapping are extracted from the reflection map and the compensated reflection map respectively and fused. The fault probability value of the candidate location is determined based on the fused feature vector, and the fault point in the optical fiber link is determined according to the fault probability value.

2. The fiber optic fault location method based on OTDR reflection images according to claim 1, characterized in that, Identifying strong reflection events from the reflection map includes: Local peak detection is performed on the reflection map to obtain candidate peak points, and the peak height and peak width of each candidate peak point are calculated respectively; wherein, the peak height is the difference between the power value of the candidate peak point and the background power value of its neighborhood, and the peak width is the difference between the corresponding distance positions when the power on the left and right sides of the candidate peak point drops to a preset proportion of the peak power; When the peak height is greater than the first threshold and the peak width is less than the second threshold, the corresponding candidate peak point is determined as a strong reflection event.

3. The fiber optic fault location method based on OTDR reflection images according to claim 2, characterized in that, Determining the corresponding occlusion range based on the strong reflection event includes: Obtain the peak position and peak height corresponding to the strong reflection event, and calculate the trailing attenuation rate within a preset observation distance range behind the peak position; The shading length is calculated based on the peak height and the trailing attenuation rate, and the interval from the peak position to the peak position plus the shading length is determined as the shading interval; wherein, the shading length is positively correlated with the peak height and negatively correlated with the trailing attenuation rate.

4. The fiber optic fault location method based on OTDR reflection images according to claim 1, characterized in that, A strong reflection trailing signal model is established for the aforementioned shielded area, including: An exponential decay function is used to model the strong reflection trail signal so that the estimated value of the strong reflection trail signal is related to the peak height and attenuation coefficient of the strong reflection event; wherein the attenuation coefficient is obtained by performing least squares fitting on the reflection signal within the shading interval; Based on the strong reflection trailing signal model, the reflection signal of the reflection map within the occlusion region is compensated and reconstructed to generate a compensated reflection map, including: The strong reflection trailing signal model is used to calculate the estimated trailing signal in the shielding area, and the estimated trailing signal is used as a compensation term to correct the reflection signal in the shielding area to obtain the compensation signal. When the compensation signal is lower than the noise baseline, the compensation signal is interpolated and corrected based on adjacent sampling points to generate the compensation reflection map.

5. The fiber optic fault location method based on OTDR reflection images according to claim 1, characterized in that, Extracting and fusing feature vectors of the reflection mapping from the reflection map and the compensated reflection map respectively, including: At least three features from peak height, peak width, local attenuation slope, second curvature, and local energy are extracted from the reflection map along the distance axis using a sliding window to form a first feature vector, and corresponding features are extracted from the compensated reflection map to form a second feature vector. The first feature vector and the second feature vector are concatenated or weighted and fused to obtain a fused feature vector; The fused feature vector is input into a machine learning classification model to output the fault probability value of the candidate location, and when the fault probability value exceeds a preset probability threshold, the corresponding candidate location is determined as a fault point.

6. A fiber optic fault location device based on OTDR reflection images, characterized in that, The device includes: The reflection pattern generation module is used to control the OTDR device to send test light pulses to the fiber optic link under test to obtain a series of reflection signals, and to generate a reflection pattern based on the series of reflection signals; The occlusion interval determination module is used to identify strong reflection events from the reflection map and determine the corresponding occlusion interval based on the strong reflection events; The compensation and reconstruction module is used to establish a strong reflection trailing signal model for the occluded area, and to compensate and reconstruct the reflection signal of the reflection map in the occluded area based on the strong reflection trailing signal model to generate a compensated reflection map. The fault location module is used to extract feature vectors of reflection mapping from the reflection map and the compensated reflection map respectively and fuse them, determine the fault probability value of the candidate location based on the fused feature vector, and determine the fault point in the optical fiber link according to the fault probability value.

7. The fiber optic fault location device based on OTDR reflection images according to claim 6, characterized in that, The occlusion zone determination module includes a strong reflection recognition unit, which is used for: Local peak detection is performed on the reflection map to obtain candidate peak points, and the peak height and peak width of each candidate peak point are calculated respectively; wherein, the peak height is the difference between the power value of the candidate peak point and the background power value of its neighborhood, and the peak width is the difference between the corresponding distance positions when the power on the left and right sides of the candidate peak point drops to a preset proportion of the peak power; When the peak height is greater than the first threshold and the peak width is less than the second threshold, the corresponding candidate peak point is determined as a strong reflection event.

8. The fiber optic fault location device based on OTDR reflection images according to claim 7, characterized in that, The occlusion interval determination module further includes an occlusion interval calculation unit, which is used for: Obtain the peak position and peak height corresponding to the strong reflection event, and calculate the trailing attenuation rate within a preset observation distance range behind the peak position; The shading length is calculated based on the peak height and the trailing attenuation rate, and the interval from the peak position to the peak position plus the shading length is determined as the shading interval; wherein, the shading length is positively correlated with the peak height and negatively correlated with the trailing attenuation rate.

9. The fiber optic fault location device based on OTDR reflection images according to claim 6, characterized in that, The compensation and reconstruction module includes: The trailing signal modeling unit is used to model the strong reflection trailing signal using an exponential decay function, so that the estimated value of the strong reflection trailing signal is related to the peak height and attenuation coefficient of the strong reflection event; wherein the attenuation coefficient is obtained by performing least squares fitting on the reflection signal within the shading interval. The compensation and correction unit is used to calculate the estimated trailing signal in the shading area using the strong reflection trailing signal model, and to use the estimated trailing signal as a compensation term to correct the reflection signal in the shading area to obtain a compensation signal. An interpolation correction unit is used to interpolate and correct the compensation signal based on adjacent sampling points when the compensation signal is lower than the noise baseline, so as to generate the compensation reflection map.

10. A computer device, characterized in that, The method includes a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements the method as described in any one of claims 1-5.

Images