An OTDR event detection method, system, and electronic device

Abstract

The application relates to the technical field of optical fiber sensing, and more particularly to an OTDR event detection method and system and an electronic device. The method comprises the following steps: performing block processing on OTDR data to obtain data blocks; the data blocks comprise feature data points and verification data points; predicted data points are obtained according to the feature data points, and abnormal data points are obtained according to the predicted data points and the verification data points; the abnormal data points carry position information; abnormal data clusters are obtained according to the abnormal data points and the OTDR data, and original data segments containing a preset data amount are obtained from the OTDR data according to the abnormal data clusters; detection data is obtained according to the original data segments; the detection data is input into an event detection model for event detection, and OTDR event detection results are obtained. The method can realize low-complexity, high-accuracy and high-flexibility identification and detection of OTDR events.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of optical fiber sensing, and more particularly to an OTDR event detection method, system and electronic device. BACKGROUND

[0002] At present, an optical time domain reflectometer (OTDR) is an optical fiber testing instrument based on the principle of optical time domain reflection. Its core function is to accurately measure the transmission characteristics of an optical fiber by transmitting a high-power optical pulse into the optical fiber and receiving the backscattered signal. The OTDR can realize non-destructive testing of an optical fiber link, can obtain key parameters such as the length, attenuation coefficient, fusion loss and bending loss of the optical fiber in real time, and can locate physical layer faults such as optical fiber breakage and connector damage. Therefore, the OTDR is widely used in the quality evaluation of optical fiber communication network construction, operation and maintenance, and data center optical fiber links.

[0003] In the use of the OTDR and the analysis of its data, an OTDR event is a node or area in an optical fiber link that causes a sudden change in signal characteristics, including but not limited to typical events such as a reflection event caused by reflection of a signal at the end face of an optical fiber, a non-reflection event caused by signal attenuation at the end of an optical fiber, and an abnormal event caused by a defect in the body of an optical fiber. The accurate detection of these OTDR events is directly related to the reliability of optical fiber transmission. Through the detection of OTDR events, the operation of the optical fiber can be controlled, and the troubleshooting and repair of the optical fiber can be achieved. Therefore, the sensitivity and accuracy of event detection are the key to ensuring the stable operation of an optical fiber communication system.

[0004] In the prior art, there are two technical paths for OTDR event detection: one is a traditional method based on threshold determination, which determines events by presetting threshold values such as reflection intensity and attenuation gradient. However, this method is prone to misjudgment in scenarios such as noise interference or optical fiber aging, and fixed threshold values cannot adapt to the detection needs of various scenarios such as multimode optical fibers and bend-insensitive optical fibers. The other is an intelligent algorithm based on signal processing, but it generally has high computational complexity and poor real-time performance. SUMMARY

[0005] The present application provides an OTDR event detection method, system and electronic device for realizing low-complexity, high-accuracy and high-flexibility identification and detection of OTDR events.

[0006] According to a first aspect of the present application, an OTDR event detection method is provided, the method comprising:

[0007] obtaining OTDR data;

[0008] The OTDR data is processed in blocks to obtain a plurality of data blocks; wherein the data block includes a plurality of feature data points and a plurality of verification data points;

[0009] According to the feature data points of each data block, the prediction data points corresponding to the verification data points are obtained, and according to the prediction data points and the verification data points of each data block, the abnormal data points are obtained; wherein the abnormal data points carry position information;

[0010] According to the abnormal data points and the corresponding position information, a plurality of abnormal data clusters are obtained from the OTDR data;

[0011] According to the abnormal data clusters, the original data segments containing a preset data amount are obtained from the OTDR data;

[0012] According to the original data segment, the corresponding detection data is obtained;

[0013] The detection data is input into an event detection model for event detection to obtain an OTDR event detection result.

[0014] It can be understood that the prediction data points corresponding to the verification data points are generated according to the feature data points, so as to obtain the abnormal data points, realize cross-validation of data, effectively reduce misjudgment caused by noise or model deviation; based on the abnormal data points, the abnormal data clusters are divided, and the detection data is obtained, which not only preserves the spatiotemporal continuity of the event, but also avoids missing detection caused by data truncation or sparse distribution; the detection data is used as input data of the event detection model, which can accurately obtain the detection data of the OTDR event in the OTDR data, thereby avoiding inputting all OTDR data into the event detection model for processing and additional redundant checking, thereby reducing the input data amount of the event detection model, saving a large amount of computing resources, and improving the processing efficiency of the event detection model.

[0015] Optionally, the OTDR data is processed in blocks to obtain a plurality of data blocks; wherein the data block includes a plurality of feature data points and a plurality of verification data points, including:

[0016] The OTDR data is divided into a plurality of data blocks according to a preset total number of data blocks and a preset moving step number; wherein the number of data points of each data block is the total number of data blocks; the first data block takes the first data point of the OTDR data as the division starting point, and the division starting points of other data blocks are separated from the division starting point of the previous data block by the moving step number;

[0017] According to a preset ratio, the feature data points and the verification data points of the data block are divided to obtain a plurality of feature data points and a plurality of verification data points corresponding to the data block.

[0018] It can be understood that the data blocks are divided by the fixed total number of data block points, and the partial overlap of adjacent data blocks is realized by combining the moving step number, which not only ensures the integrity of the local features, but also balances the computing resources and the signal continuity demand through the step adjustment. According to the preset proportion, the feature data points and the verification data points are obtained, which can adapt to the OTDR data of different sampling rates, fiber lengths and event densities, so as to reduce the data processing complexity while ensuring the detection accuracy, and significantly optimize the fineness and efficiency of the OTDR data processing.

[0019] Optionally, the method further comprises:

[0020] According to the feature data points, a plurality of prediction data points are obtained by data prediction, and one prediction data point corresponds to one verification data point;

[0021] The prediction data points are subtracted from the corresponding verification data points to obtain difference values;

[0022] The maximum difference value in the difference values is obtained;

[0023] According to the difference values corresponding to the verification data points and the maximum difference value, an abnormal data point is determined.

[0024] It can be understood that the prediction data points and the verification data points form a one-to-one correspondence, which ensures the strict matching of the data space positions and avoids the misjudgment caused by data mispositioning in the traditional method; by calculating the difference values of the prediction data points and the verification data points and introducing the maximum difference value as another data basis for judgment, the abnormal situation of the verification data points is accurately judged, so as to reduce the false alarm rate, improve the detection sensitivity of weak faults, and significantly improve the accuracy of abnormal situation detection.

[0025] Optionally, the method further comprises:

[0026] A first difference threshold value and a second difference threshold value are preset, and the second difference threshold value is greater than the first difference threshold value;

[0027] It is judged whether the difference value corresponding to the verification data point exceeds the first difference threshold value and whether the maximum difference value exceeds the second difference threshold value, and if so, the verification data point is taken as an abnormal data point.

[0028] It can be understood that the preset first difference threshold is taken as the judgment basis of individual anomaly, and the preset second difference threshold is taken as the intensity threshold of global anomaly, realizing double checking of local deviation and overall anomaly degree, effectively excluding isolated over-limit phenomenon caused by local noise or slight disturbance, and ensuring that only when individual significant deviation and overall anomaly exist simultaneously in the data block, the setting of anomaly is triggered, finally realizing double improvement of anomaly positioning accuracy and detection confidence, thereby significantly improving the identification accuracy and robustness of verifying data point anomaly.

[0029] Optionally, the obtaining of the plurality of abnormal data clusters from the OTDR data according to the abnormal data points and the corresponding position information comprises:

[0030] According to the position information of the two adjacent abnormal data points, two OTDR data points with the same position information in the OTDR data are obtained.

[0031] The number of OTDR data points between the two OTDR data points is obtained.

[0032] If the number of OTDR data points is less than or equal to a preset first clustering threshold, the corresponding two OTDR data points and the OTDR data points between the two OTDR data points are taken as an abnormal data cluster.

[0033] It can be understood that, taking the distance between adjacent abnormal data points as the judgment basis, when the number of OTDR data points between two points is less than the preset threshold, the intermediate data points are merged as an abnormal data cluster, which can consider that the abnormal data corresponding to the abnormal event caused in the optical fiber often presents continuous distribution, while noise or isolated interference presents sparse discrete points, so the spatial continuity feature of optical fiber fault is ingeniously utilized to obtain the abnormal data cluster corresponding to the situation; through adjustment of the first clustering threshold, the same physical event can be combined into a complete data cluster to avoid fragmentation caused by data sampling interval, and the positioning accuracy of the optical fiber abnormal position can be improved, and the correlation analysis efficiency of the abnormal event is significantly improved.

[0034] Optionally, the obtaining of the original data segment containing a preset data amount from the OTDR data according to the abnormal data cluster comprises:

[0035] The number of data points of the abnormal data cluster is obtained.

[0036] If the number of data points of the abnormal data cluster does not exceed a preset second clustering threshold, the first OTDR data point of the abnormal data cluster is taken as a division node.

[0037] If the number of data points of the abnormal data cluster exceeds the second clustering threshold but does not exceed a preset third clustering threshold, a plurality of division nodes of the abnormal data cluster are obtained at a preset data point interval;

[0038] If the number of data points of the abnormal data cluster exceeds the third clustering threshold, a preset third number of OTDR data points are extracted from the end of the corresponding abnormal data cluster to update the corresponding abnormal data cluster, and a plurality of division nodes are obtained in the updated abnormal data cluster at the data point interval;

[0039] According to the position information corresponding to the division nodes, a raw data segment containing a preset data amount is obtained from the OTDR data.

[0040] It can be understood that if the abnormal data cluster is small in size, the first OTDR data point thereof is directly obtained as a division node; if the abnormal data cluster is medium in size, a plurality of division nodes thereof are obtained at a preset data point interval; and if the abnormal data cluster is large in size, a core abnormal data of a third number of data points is retained by starting to cut data from the end, and a plurality of division nodes are obtained, which not only eliminates redundant and useless data but also retains the key features of the abnormal data cluster, thereby avoiding waste of subsequent processing power due to the large size of the abnormal data cluster.

[0041] Optionally, the detection data includes power domain detection data and voltage domain detection data.

[0042] The detection data is input into an event detection model for event detection to obtain an OTDR event detection result, including:

[0043] The power domain detection data and the voltage domain detection data are respectively input into an event detection model for event recognition to obtain a plurality of power domain abnormal events and a plurality of voltage domain abnormal events.

[0044] Misjudgment events are obtained by checking the power domain abnormal events.

[0045] The misjudgment events are deleted from the power domain abnormal events to obtain updated power domain abnormal events.

[0046] According to the voltage domain abnormal events and the updated power domain abnormal events, abnormal event information in the OTDR data is obtained.

[0047] The OTDR event detection result is obtained according to the abnormal event information.

[0048] It can be understood that the OTDR power domain data and the OTDR voltage domain data are synchronously collected, independently processed and event-identified, and accurately determined through cross-domain cross verification; the OTDR power domain data is good at capturing strong reflection events of the optical fiber structure, but it is easily interfered by environmental noise, leading to misjudgment; the OTDR voltage domain data is sensitive to electrical signal fluctuations and can assist in identifying device abnormalities or external interference. Through misjudgment investigation of the power domain abnormal event and logical verification combined with the voltage domain abnormal event, false alarms caused by device transient disturbance or environmental vibration are effectively filtered out; at the same time, independent processing of the dual-domain data retains the respective characteristic advantages, so that the model can distinguish between gradual faults such as microbending and temperature change and sudden breaking events.

[0049] Optionally, the misjudgment investigation on the power domain abnormal event to obtain a misjudgment event specifically includes:

[0050] Obtaining an optical fiber position where the power domain abnormal event exists and the voltage domain abnormal event does not exist;

[0051] From the OTDR data, an OTDR voltage domain data segment of a preset length centered on the optical fiber position is extracted;

[0052] Obtaining a signal-to-noise ratio decay value of the OTDR voltage domain data segment, if the signal-to-noise ratio decay value does not exceed a preset decay threshold, determining the power domain abnormal event corresponding to the optical fiber position as a misjudgment event; and / or, obtaining a jitter index of the OTDR voltage domain data segment, if the jitter index exceeds a preset jitter threshold, determining the power domain abnormal event corresponding to the optical fiber position as a misjudgment event.

[0053] It can be understood that false alarms caused by poor optical fiber quality are filtered out through the signal-to-noise ratio decay value, when the signal-to-noise ratio does not reach the decay threshold, it indicates that the power abnormality may be caused by noise interference rather than real optical fiber damage; secondly, the jitter index is used to identify transient power fluctuations caused by environmental disturbance, if the jitter exceeds the standard, it is determined as dynamic interference rather than static failure. This double-condition serial determination method not only retains the high sensitivity of the power domain data to the changes in the optical fiber structure, but also realizes misjudgment investigation through the physical layer state feedback of the voltage domain, so as to reduce the false positive rate of event detection and provide a more reliable abnormality determination basis for event detection.

[0054] According to a second aspect of the present application, an OTDR event detection system is provided, the system comprising:

[0055] An OTDR data acquisition module for acquiring OTDR data;

[0056] A data block acquisition module for block processing the OTDR data to obtain a plurality of data blocks; wherein the data block includes a plurality of feature data points and a plurality of verification data points;

[0057] anomaly obtaining module, configured to obtain a predicted data point corresponding to the verification data point according to a feature data point of each data block, and obtain an anomaly data point according to the predicted data point and the verification data point of each data block, wherein the anomaly data point carries position information;

[0058] clustering module, configured to obtain a plurality of anomaly data clusters from the OTDR data according to the anomaly data points and corresponding position information;

[0059] original data segment obtaining module, configured to obtain an original data segment containing a preset data amount from the OTDR data according to the anomaly data cluster;

[0060] detection data obtaining module, configured to obtain corresponding detection data according to the original data segment;

[0061] detection result obtaining module, configured to input the detection data into an event detection model to perform event detection, and obtain an OTDR event detection result.

[0062] According to a third aspect of the present application, an electronic device is provided, comprising:

[0063] a memory, configured to store one or more computer programs;

[0064] a processor, when the one or more computer programs are executed by the processor, implements the OTDR event detection method of the first aspect.

[0065] According to a fourth aspect of the present application, a computer readable storage medium is provided, the computer readable storage medium stores computer instructions, and the computer instructions are used to make the processor execute the OTDR event detection method of the first aspect.

[0066] Based on any one of the above aspects, the OTDR event detection method, system, electronic device and storage medium provided by the embodiments of the present application can achieve the following benefits:

[0067] Through the fine data processing process, the accuracy of the data is ensured: through the blocking processing of the OTDR data and the binding of the position information, the fine-grained spatial positioning of the OTDR data is realized. Each data block contains a check structure of feature data points and verification data points, the anomaly data points are determined based on the predicted data points, which can effectively eliminate noise interference and ensure the accurate division of the anomaly data cluster; the integration strategy of the original data segment and the anomaly data cluster is adopted, the original data features and position correlation of the OTDR data are retained, the event feature fragmentation problem caused by the traditional segmentation processing is avoided, and the reliability of the subsequent analysis is improved from the data preprocessing level.

[0068] • Improve the accuracy of event detection through multi-domain data fusion: Obtain power domain detection data and voltage domain data through OTDR data. The power domain detection data focuses on the analysis of fiber attenuation characteristics, and the voltage domain detection data focuses on the characteristics of reflected signals. Through the complementary identification model of the two physical domains, the ability to distinguish events such as fiber microbending, breakage, and connector failure is significantly improved.

[0069] • Optimize the credibility of event detection results by correcting misjudgment events: Misjudgment events are investigated for abnormal events in the power domain. Dynamic filtering of pseudo-abnormalities caused by normal attenuation, jitter, and other non-fault factors is performed to automatically separate misjudgment events. This breaks the limitation of single-domain data being easily disturbed by environmental noise, realizes multi-dimensional feature cross-validation, reduces the false alarm rate and the missed detection rate, and effectively solves the pain points of easy mis-triggering of traditional OTDR single-parameter analysis. BRIEF DESCRIPTION OF DRAWINGS

[0070] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0071] Figure 1 A schematic application scenario diagram of an OTDR event detection method provided by the present embodiment.

[0072] Figure 2 A flowchart of an OTDR event detection method provided by the present embodiment.

[0073] Figure 3 A flowchart of obtaining a data block provided by the present embodiment.

[0074] Figure 4 A flowchart of obtaining an abnormal data point provided by the present embodiment.

[0075] Figure 5 A flowchart of obtaining an abnormal data cluster provided by the present embodiment.

[0076] Figure 6 A flowchart of obtaining an original data segment provided by the present embodiment.

[0077] Figure 7 A flowchart of obtaining an OTDR event detection result provided by the present embodiment.

[0078] Figure 8 A flowchart of obtaining a misjudgment event provided by the present embodiment.

[0079] Figure 9A functional module schematic diagram of an OTDR event detection system provided by the embodiment.

[0080] Figure 10 A structural schematic diagram of an electronic device provided by the embodiment. DETAILED DESCRIPTION

[0081] The drawings of the present application are only used for illustrative description, and cannot be understood as a limitation of the present application. In order to better illustrate the following embodiments, some components in the drawings will be omitted, enlarged or reduced, and the size of the actual product is not represented; it is understandable for those skilled in the art that some well-known structures and their descriptions in the drawings can be omitted.

[0082] In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the scope of protection of the present application.

[0083] It should be noted that the terms "first", "second" and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to the process, method, product or device.

[0084] OTDR (Optical Time-Domain Reflectometer) is a kind of precise test instrument specially designed for optical fiber communication network, its core principle is to transmit high-power optical pulse to optical fiber, and receive backscattered light and reflected light signals generated by optical fiber, and then realize comprehensive diagnosis of optical fiber link. OTDR device can measure the physical parameters such as optical fiber length, transmission loss and joint loss; can accurately locate the fault points such as optical fiber fracture and excessive bending; can evaluate the overall health status of optical fiber link, and provide data support for construction acceptance and daily maintenance.

[0085] OTDR event refers to an abnormal feature point in the optical fiber link identified by the OTDR test curve obtained by the OTDR device. Common types include fiber absorption events, core breakage, bending events, scattering events, and reflection events. These events correspond to different physical phenomena of the optical fiber, thereby enabling rapid positioning of the abnormal point of the optical fiber; OTDR events can also quantitatively evaluate the degree of performance degradation of the optical fiber link, providing a basis for preventive maintenance.

[0086] The current mainstream OTDR event detection method mainly relies on manual analysis of the response curve or simple slope method algorithm. These existing detection methods have significant limitations: high noise sensitivity, low signal-to-noise ratio of collected data, leading to low test efficiency; blind zone effect in the optical fiber may limit the detection accuracy of the short-distance event in the optical fiber, especially the OTDR event blind zone may cover the continuous reflection points within a short distance; contradiction between dynamic range and pulse width, long-distance test requires large pulse width but will expand the data volume and data resolution, affecting the multi-event resolution capability; manual interpretation is easily affected by subjective factors, which may lead to false positives or missed detection. Although emerging technologies attempt to introduce machine learning algorithms to optimize threshold selection, in complex optical fiber environments, the sensitivity decline problem caused by high loss cannot be solved, and the training data and parameters required by machine learning are too many, greatly increasing the complexity of detection.

[0087] The embodiment provides a technical solution that can solve the above problems. The specific embodiments of the application will be described in detail below with reference to the accompanying drawings.

[0088] Exemplarily, an application scenario of an OTDR event detection method provided by the embodiment of the application is shown. As shown in the figure, Figure 1 The application scenario at least includes a server 100 and a terminal 200 that can communicate with the server 100. The server 100 has data processing and data analysis functions; the terminal 200 has data processing, data analysis, and data display functions.

[0089] It can be understood that the server 100 can be an independent electronic device or a cluster composed of multiple electronic devices; the terminal 200 can be a smartphone terminal, a personal computer, a tablet computer, a vehicle-mounted terminal, etc., but is not limited thereto.

[0090] In an implementable manner, the server 100 and the terminal 200 can respectively execute the OTDR event detection method provided by the embodiment of the application, or alternatively, the OTDR event detection method provided by the embodiment of the application is partially executed in the server 100 and partially executed in the terminal 200.

[0091] As shown in the figure, Figure 2As shown, the embodiment provides an OTDR event detection method, which can be divided into the following steps:

[0092] S100, acquiring OTDR data;

[0093] In the embodiment, the OTDR data is obtained by collecting the optical fiber to be tested by the OTDR device.

[0094] Specifically, the OTDR data acquisition includes acquiring OTDR power domain data and OTDR voltage domain data.

[0095] In the embodiment, the corresponding power domain data needs to be extracted from the acquired OTDR data, and the corresponding voltage domain data is acquired according to the power domain data. Preferably, the corresponding voltage domain data is acquired by the following formula:

[0096]

[0097] wherein, is the voltage domain data, is the power domain data, and A and B are conversion parameters. Preferably, A can be 10, and B can be 5. The values of A and B can be adjusted according to actual conditions.

[0098] Specifically, in the embodiment, the OTDR power domain data and the OTDR voltage domain data need to be processed by steps S200 to S600 to obtain corresponding power domain detection data and voltage domain detection data.

[0099] In the embodiment, processing the OTDR data will correspondingly obtain different intermediate data, including data points, data blocks, data clusters, and data segments. It can be understood that the OTDR data collected by the OTDR device is generally a plurality of discrete data points, each data point containing corresponding position information; a plurality of adjacent data points form a data block, which is a basic unit for subsequent local analysis and prediction; a plurality of data points with similar characteristics obtained by prediction can be classified as a data cluster, which is used for subsequent identification of abnormalities. Preferably, the data points with similar characteristics can be a plurality of data points judged as abnormal; a plurality of adjacent data points obtained in the OTDR data according to a specific division identifier can form a data segment, which provides necessary data basis for subsequent event detection.

[0100] S200, block processing the OTDR data to obtain a plurality of data blocks; wherein the data blocks include a plurality of feature data points and a plurality of verification data points;

[0101] In the embodiment, the OTDR data is processed in blocks, the large and continuous OTDR data can be divided into several data blocks, and the several data blocks can be processed synchronously in batches, so that the efficiency of processing the OTDR data is improved.

[0102] It can be understood that the division of the feature data points and the verification data points for each data block can provide a data basis for subsequent prediction of the OTDR data.

[0103] Specifically, as shown in Figure 3 The OTDR data is processed in blocks to obtain several data blocks; wherein the data block includes several feature data points and several verification data points, and includes the following steps:

[0104] S210, the OTDR data is divided into blocks according to a preset total number of data blocks and a preset moving step number, to obtain several data blocks; wherein the number of data points of each data block is the total number of data blocks; the first data block takes the first data point of the OTDR data as the division starting point, and the division starting point of other data blocks is separated from the division starting point of the previous data block by the moving step number;

[0105] It can be understood that the OTDR data collected by the OTDR device is generally several discrete OTDR data points, and the several discrete OTDR data points need to be identified in position, so that the processing of each OTDR data point is more clear and orderly, and the confusion and corresponding errors in processing of different OTDR data points are avoided. In the embodiment, the position information is configured for each OTDR data point to complete the identification of the OTDR data point in position, and preferably, the position information is the sequence index number corresponding to each OTDR data point.

[0106] In the embodiment, the total number of data blocks can limit the number of OTDR data points contained in the data block, so that the size of each data block is the same, and the input data of the standard format is improved for prediction processing. The moving step number defines the division rule of the data block, and by moving the OTDR data point of the moving step number, the division of the next data block is started, which can capture the gradual change characteristics in the continuous data block and avoid the sudden truncation error caused by the traditional non-overlapping block.

[0107] Exemplarily, the total number of data blocks can be set to 12, and the moving step number can be set to 2, and the specific value can be adjusted according to the actual situation.

[0108] In the embodiment, the first data block is divided from the first data point of the OTDR data, and the OTDR data points of the total data block points are divided into the first data block; the division starting point of the next data block is the OTDR data point of the first data point of the OTDR data moving the moving step points, and the OTDR data points of the total data block points are divided into the next data block. In this way, a plurality of data blocks of the same size and with repeated data points are obtained.

[0109] For example, if the order index number of the OTDR data is set to start from 0, the first divided data block is the OTDR data points containing the order index numbers 【0-11】; and the division starting point of the second data block is the OTDR data point with the order index number 2, and the second divided data block is the OTDR data points containing the order index numbers 【2-13】. In this way, until all the OTDR data points of the OTDR data are divided.

[0110] In the embodiment, the proportion is used to limit the number of feature data points and verification data points, so that the feature data points used for prediction of each data block are the same number, and the predicted prediction data points can also correspond to the specified number of verification data points one by one, improving the efficiency of data recognition and prediction processing efficiency.

[0111] In the embodiment, the proportion is used to limit the number of feature data points and verification data points, so that the feature data points used for prediction of each data block are the same number, and the predicted prediction data points can also correspond to the specified number of verification data points one by one, improving the efficiency of data recognition and prediction processing efficiency.

[0112] For example, the proportion can be 2:1, and if the total data block points contained in the limited data block are 12, the feature data points can be set to 8, and the verification data points can be set to 4. The specific value can be adjusted according to the actual situation.

[0113] Preferably, the number of feature data points can also be preset, and the data block is divided based on the number of feature data points; for example, the number of feature data points is preset to 8, and if the total data block points contained in the limited data block are 12, the feature data points can be set to 8, and the remaining OTDR data points of the data block are divided into verification data points. The specific value can be adjusted according to the actual situation.

[0114] Preferably, the feature data points and the verification data points are divided in the data block, and the data block carries position information, so the feature data points and the verification data points also carry corresponding position information.

[0115] Exemplarily, in the first data block, the OTDR data points with the sequential index numbers of 【0-11】 are contained, the OTDR data points with the sequential index numbers of 【0-7】 can be divided into the feature data points, and the OTDR data points with the sequential index numbers of 【8-11】 can be divided into the verification data points. The other data blocks are divided according to the above uniform rules, and details are not described herein.

[0116] S300, obtaining the predicted data points corresponding to the verification data points according to the feature data points of each data block, and obtaining the abnormal data points according to the predicted data points and the verification data points of each data block; wherein the abnormal data points carry position information;

[0117] In the embodiment, the feature data points are predicted to be the data points corresponding to the tail of the same data block as the feature data points, and the predicted data points obtained by the prediction are compared with the verification data points to determine whether the verification data points are abnormal data points. It can be understood that the prediction algorithm can obtain the development curve of the subsequent feature data points according to the characteristics and historical experience of the optical fiber, and the prediction data points are obtained by the prediction data points, that is, the prediction data points represent the predicted optical fiber based on the normal operation of the feature data points. If the predicted data points and the real verification data points have a large deviation, it means that the optical fiber may have a fault at this position, and subsequent event processing is required at this position. Preferably, based on the linear regression algorithm, the feature data points are taken as input data to predict the OTDR data points corresponding to the verification data points to obtain the predicted data points, and then the abnormal conditions of the OTDR data points corresponding to the verification data points are obtained according to the comparison between the predicted data points and the verification data points.

[0118] Specifically, as shown in Figure 4 the feature data points of each data block, and obtaining the abnormal data points according to the predicted data points and the verification data points of each data block, comprises:

[0119] S310, obtaining a plurality of predicted data points according to data prediction of the feature data points, and one predicted data point corresponding to one verification data point;

[0120] In the embodiment, the predicted data points obtained need to be one-to-one corresponding to the OTDR data of the verification data points, so that the subsequent pairwise comparison can be accurately completed. Preferably, the verification data points carry position information, and the predicted data points also carry corresponding position information.

[0121] S320, obtaining the difference between the predicted data points and the corresponding verification data points to obtain the difference value;

[0122] S330, obtaining the maximum difference value in the difference value;

[0123] In the embodiment, the difference between the single predicted data point and the single verification data point can reflect the operation of the single fiber position corresponding to the data point, and the maximum difference obtained can reflect the overall operation of the fiber region corresponding to the verification data point. Taking the overall operation of the fiber region as one of the conditions for fault judgment can not only take the data deviation of the single fiber position as the only judgment point of the fault, but also avoid the error of the single fiber position data jitter, thereby reducing the misjudgment rate of the abnormality.

[0124] S340, determining an abnormal data point according to the difference corresponding to the verification data point and the maximum difference;

[0125] Preferably, the abnormal data points are labeled as abnormal, and the other verification data points are labeled as normal, facilitating subsequent processing and statistics of the data.

[0126] In the embodiment, both the difference between the single predicted data point and the single verification data point and the maximum difference are used as the discrimination condition of the abnormality, and the corresponding abnormal data point is obtained, which is labeled as abnormal, so that the abnormal data point can be located according to the abnormal label in the subsequent process, improving the accuracy of data processing. Understandably, the verification data in the plurality of data blocks can be predicted, wherein the adjacent two data blocks contain part of the repeated data points. The trend change of each OTDR data point in the verification data can be predicted based on the related pre-data, and the cross-validation method can filter accidental noise interference and improve the reliability of abnormal detection.

[0127] In a new implementation, the abnormal label can be represented by a label value corresponding to each OTDR data point, wherein the label value can be set as 1 if the corresponding OTDR data point is abnormal, and 0 if the corresponding OTDR data point is normal. Before the step of judging whether the OTDR data point is abnormal, the label value of all OTDR data points can be set to 1, i.e. all OTDR data points are considered to be abnormal, and the label value of the OTDR data point is updated in the subsequent abnormal judgment, improving the processing efficiency of the label setting.

[0128] Specifically, the determination of the abnormal data point according to the difference corresponding to the verification data point and the maximum difference comprises:

[0129] A first difference threshold and a second difference threshold are preset, and the second difference threshold is greater than the first difference threshold;

[0130] In the embodiment, the first difference threshold is used to measure the deviation of the difference corresponding to the verification data point, and the second difference threshold is used to measure the deviation of the maximum difference. The first difference threshold and the maximum difference can be adjusted as appropriate according to actual conditions.

[0131] It is judged whether the difference corresponding to the verification data point exceeds the first difference threshold and whether the maximum difference exceeds the second difference threshold. If yes, the verification data point is regarded as an abnormal data point.

[0132] In the embodiment, the verification data point is regarded as an abnormal data point only when two conditions are met, i.e., the difference corresponding to the verification data point exceeds the first difference threshold and the maximum difference exceeds the second difference threshold. In a new implementation, the label value of the abnormal data point is kept as 1. If the two conditions are not met, the verification data point is set to a normal label. In a new implementation, the label value of the verification data point is updated to 0.

[0133] Preferably, the first data block is the characteristic data of the OTDR data points with sequential index numbers 0-7. In the prediction process, the first 8 OTDR data points of the first data block are not divided into verification data points. Considering that the front data generally reflects the normal reflection image caused by the head of the optical fiber, the normal reflection image is not considered in the event detection of the present application, and the front data generally corresponds to the characteristic data points of the first data block, the characteristic data points of the first data block, i.e., the OTDR data points with sequential index numbers 0-7, are set to a normal label in the abnormality judgment. In a new implementation, the label value corresponding to the OTDR data points with sequential index numbers 0-7 is updated to 0, so that the processed data is consistent with the normal operation of the optical fiber.

[0134] Preferably, the slope and the intercept corresponding to the verification data points as a whole can also be obtained according to the characteristic data points of each data block. The slope is used to assist in judging and predicting the attenuation of the verification data points, and the intercept is used to assist in judging and predicting the noise level of the verification data points. If the attenuation is consistent with the normal attenuation level and / or the noise level is close to the general noise floor range of the OTDR equipment, and the result of the difference comparison is considered, the verification data points are set to a normal label. Preferably, the noise floor range fluctuates by 5db. The noise floor range can be adjusted as appropriate according to the operation of the OTDR equipment and the optical fiber.

[0135] S400, obtaining a plurality of abnormal data clusters from the OTDR data according to the abnormal data points and the corresponding position information;

[0136] The abnormal data points carry position information, and the obtained abnormal data clusters also carry position information.

[0137] In the embodiment, the two adjacent abnormal data points can be in the same data block or in different data blocks. Therefore, the event identification processing according to the data blocks of the uniform length cannot obtain the abnormal data related to the entire abnormal OTDR event, and thus the OTDR data needs to be re-divided according to the abnormal data points in the data blocks, and the abnormal data clusters are divided in the form of abnormal data clusters, so as to integrate the data possibly belonging to one OTDR event into one cluster as much as possible, thereby reducing the repeated identification of the abnormal OTDR event.

[0138] Specifically, as shown in Figure 5 The abnormal data clusters are obtained from the OTDR data according to the abnormal data points and the corresponding position information, and the method comprises the following steps of:

[0139] S410, two OTDR data points with the same position information are obtained from the OTDR data according to the position information of the two adjacent abnormal data points.

[0140] In the embodiment, the position information of the two adjacent abnormal data points can also be the sequence index number, and the positions of the corresponding OTDR data points in the OTDR data are located. At this time, one or more OTDR data points can exist between the two adjacent abnormal data points, and if the labels are set, the OTDR data points have normal labels or the label values are 0. In this case, the step S420 and the following steps need to be completed.

[0141] The two abnormal data points can be directly connected in sequence without OTDR data points therebetween, and in this case, the two abnormal data points do not need to be processed or clustered.

[0142] S420, the number of OTDR data points between the two OTDR data points is obtained.

[0143] S430, if the number of OTDR data points is less than or equal to a preset first clustering threshold, the two corresponding OTDR data points and the OTDR data points between the two OTDR data points are taken as one abnormal data cluster.

[0144] Preferably, an abnormal label is set for all the OTDR data points in the abnormal data cluster.

[0145] In the embodiment, when the number of OTDR data points is less than or equal to the preset first clustering threshold, it can be considered that the abnormal phenomena of the two OTDR data points are possibly caused by the same abnormal OTDR event, and the OTDR data points between the two OTDR data points can be used as data basis for analyzing the abnormal OTDR event, so that the corresponding two OTDR data points and the OTDR data points between the corresponding two OTDR data points are required to be taken as an abnormal data cluster, and all OTDR data points in the abnormal data cluster are required to be set with an abnormal label as a discrimination data of a subsequent abnormal OTDR event.

[0146] In the implementation process, the position information of the data block, the feature data point, the verification data point, the prediction data point, the abnormal data point and the abnormal data cluster is the position information of the data point of the data block, the feature data point, the verification data point, the prediction data point, the abnormal data point and the abnormal data cluster in the OTDR data, so that the data point of the data block, the feature data point, the verification data point, the prediction data point, the abnormal data point and the abnormal data cluster are one-to-one corresponding to the OTDR data, ensuring the consistency of data position and improving the efficiency of data positioning.

[0147] S500, acquiring an original data segment containing a preset data amount from the OTDR data corresponding to the abnormal data cluster;

[0148] In the embodiment, acquiring the original data segment corresponding to the abnormal data cluster can provide original comparison of OTDR data for identification of the OTDR event, and can further verify the occurrence of the OTDR event and the continuity and trend of the OTDR event in the OTDR data, so as to realize multiple data comparison of the OTDR event.

[0149] Specifically, as shown in the figure, Figure 6 The step of acquiring the original data segment containing a preset data amount from the OTDR data corresponding to the abnormal data cluster includes the following steps:

[0150] S510, acquiring the number of data points of the abnormal data cluster;

[0151] S520, if the number of data points of the abnormal data cluster does not exceed the preset second clustering threshold, the first OTDR data point of the abnormal data cluster is taken as a division node;

[0152] S530, if the number of data points of the abnormal data cluster exceeds the second clustering threshold but does not exceed a preset third clustering threshold, a plurality of division nodes of the abnormal data cluster are acquired at a preset data point interval;

[0153] S540, if the number of data points of the abnormal data cluster exceeds the third clustering threshold, extracting OTDR data points of a preset third data point number from the end of the corresponding abnormal data cluster to update the corresponding abnormal data cluster, and obtaining a plurality of division nodes in the updated abnormal data cluster at the data point interval.

[0154] Exemplarily, the first OTDR data point in the abnormal data cluster is the first division node, and the OTDR data point at a preset data point interval from the first OTDR data point is the second division node. In this way, all division nodes of the abnormal data cluster are obtained.

[0155] S550, according to the position information corresponding to the division node, obtaining a raw data segment containing a preset data amount from the OTDR data.

[0156] In this embodiment, according to the position information corresponding to the division node, an OTDR data segment starting from the division start point is obtained from the OTDR data, wherein the OTDR data segment includes OTDR data points of a preset data amount, and the OTDR data segment is taken as a raw data segment.

[0157] In this embodiment, if the number of data points of the abnormal data cluster does not exceed the second clustering threshold, it means that the abnormal data cluster contains fewer OTDR data points. Among these OTDR data points, at most one OTDR event may occur according to experience, so the first OTDR data point of the abnormal data cluster is taken as a division node, and the corresponding raw data segment is obtained based on the division node.

[0158] In this embodiment, for the abnormal data cluster whose number of data points exceeds the second clustering threshold but does not exceed the preset third clustering threshold, the OTDR data points contained are more. A simple detection of the abnormal data cluster may miss some features corresponding to OTDR events. There may be multiple OTDR events in the abnormal data cluster, so multiple division nodes need to be obtained, multiple raw data segments are obtained based on the multiple division nodes, OTDR event features existing are captured as much as possible, corresponding OTDR events are identified, and the influence of the judgment error of the abnormal OTDR event is eliminated.

[0159] In the embodiment, the third clustering threshold is used to identify abnormal data clusters with too much noise or even all noise and / or poor quality of the fiber, and the front data of these abnormal data clusters may have too much noise and have little reference value, so the OTDR data points of a preset third data point number at the end of the abnormal data cluster are taken as update data to update the corresponding abnormal data cluster, so as to avoid processing too much redundant data and wasting unnecessary processing resources. Preferably, the second clustering threshold can be 40, which can be adjusted according to actual conditions.

[0160] Preferably, the third clustering threshold can be 1000 or other values, and the data point interval can be 40 or 60 or other values. The third clustering threshold and the preset data point interval can be adjusted according to actual conditions.

[0161] Preferably, the second clustering threshold is less than the third clustering threshold, and the data point interval is less than or equal to the second clustering threshold.

[0162] Preferably, if the abnormal data cluster is the first data of the OTDR data, the first division node can be the OTDR data point corresponding to the left shift of 8 OTDR data points of the first OTDR data point of the abnormal data cluster, so as to exclude the influence of data abnormal offset; wherein the number of left shift data points corresponds to the number of data points corresponding to the fixed setting of the normal label.

[0163] And / or, if the number of data points of the abnormal data cluster exceeds the preset third clustering threshold, the first division node can be the OTDR data point corresponding to the right shift of 3 OTDR data points of the first OTDR data point of the abnormal data cluster, so as to enhance the data characteristics of the original data segment and prevent overfitting.

[0164] S600, obtaining corresponding detection data according to the original data segment;

[0165] In the embodiment, the acquired detection data includes acquired power domain detection data and voltage domain detection data, and the detection data can be used as input data of a subsequent event detection model to identify and detect OTDR events. Preferably, the detection data includes an original data segment, a unique identifier of the original data segment, position information corresponding to the original data segment, and basic parameters corresponding to the original data segment, wherein the unique identifier is assigned by the event detection model, and the basic parameters are acquired by the original data segment. Specifically, the basic parameters can include a distance between any two OTDR data points corresponding to the original data segment, a wavelength of the OTDR data point, and reflectivity, etc. The basic parameters can be used to calculate the actual length of the optical fiber, the characteristics of the pulse light in the optical fiber transmission, and other basic information, to provide more basic information about the optical fiber for identifying OTDR events, and to improve the identification accuracy of identifying OTDR events.

[0166] S700, inputting the detection data into the event detection model for event detection to obtain an OTDR event detection result.

[0167] In the embodiment, the event detection model uses a CNN (Convolutional Neural Network) model, specifically a lightweight one-dimensional convolutional neural network, which can reshape each original data segment into a corresponding model processing tensor, and obtain corresponding features by matching the basic parameters corresponding to the original data segment, to accurately identify different types of OTDR events.

[0168] The model has a simple structure, and is provided with two convolutional layers, two pooling layers, and three fully connected layers. The convolutional layers can extract data features such as data mutations and periodic changes corresponding to the original data segment, and can extract features from low to high level layer by layer, and complete corresponding data padding during the extraction process to maintain the length of the intermediate data and maximize the extraction of data features.

[0169] The pooling layer can reduce the length of the data by a preset proportion of the original data, and the preset proportion can be 1 / 4, so as to focus on the main data features and reduce the risk of overfitting.

[0170] The fully connected layer can reduce the dimension of the data features obtained by convolution and pooling, so as to output corresponding OTDR event identification results.

[0171] The parameter amount used by the model can be determined according to the length of the input data. For example, if the length of the input data is 256, about 100,000 parameters are needed, which is a medium level compared with the parameter amount of other models. The model has great flexibility, supports any input data length that is a multiple of 4, and adapts to the processing of the model pooling layer. The types of OTDR event classification can be adjusted appropriately to support the identification and classification of different OTDR events. The application also allows the fusion of additional features, including but not limited to other basic information of the optical fiber, so that the model can have a comprehensive understanding of the optical fiber and improve the identification accuracy of OTDR events in the optical fiber.

[0172] Preferably, the Dropout regularization technique is introduced into the model, specifically, Dropout is added after the first fully connected layer, which can prevent the model from overfitting and improve the robustness of the model, which is suitable for OTDR data processing scenarios containing small noise and / or containing more noise; ReLU activation function is used in the model, specifically, ReLU activation function processing is designed after the first fully connected layer and the second fully connected layer, which can improve the nonlinear expression ability of the model and make the OTDR data processing more accurate.

[0173] Specifically, the detection data includes power domain detection data and voltage domain detection data; as Figure 7 As shown, the step of inputting the detection data into the event detection model for event detection to obtain an OTDR event detection result can include the following steps:

[0174] S710, input the power domain detection data and the voltage domain detection data into the event detection model for event recognition, to obtain a plurality of power domain abnormal events and a plurality of voltage domain abnormal events;

[0175] In this embodiment, the power domain detection data and the voltage domain detection data are input into the event detection model for event recognition, to obtain power domain recognition results and voltage domain recognition results. Preferably, the power domain recognition results and the voltage domain recognition results are normalized and converted into corresponding probability values by SoftMax function, and each probability value is processed by argmax function to obtain the OTDR event corresponding to each probability value. The types of the OTDR events include but are not limited to abnormal events, reflection events, and attenuation events, and these OTDR events have corresponding event labels to distinguish the types of the OTDR events corresponding to the positions of the optical fiber. By identifying these OTDR events, the running conditions of each position on the optical fiber can be identified, and the processing efficiency of subsequent optical fiber maintenance and operation can be improved.

[0176] Preferably, after identifying the OTDR event and the corresponding type, the event label corresponding to the OTDR event and the corresponding probability value need to be attached to the corresponding power domain detection data or voltage domain detection data, and serialized into JSON format for output, obtaining a plurality of power domain abnormal events and a plurality of voltage domain abnormal events; if no valid data of the OTDR event is obtained in the detection data, that is, no OTDR event is detected, an empty list is returned to indicate that the detection data has no power domain abnormal event or a plurality of voltage domain abnormal events.

[0177] S720, misjudgment of the power domain abnormal event is investigated to obtain a misjudgment event;

[0178] In the embodiment, the sensitivity of the possible model to the power domain data may cause misjudgment of the power domain abnormal event. The misjudgment may provide false information for subsequent optical fiber abnormality investigation and repair, so that unnecessary operation and maintenance work may be generated. Therefore, the misjudgment event needs to be further investigated and deleted from the power domain abnormal event to improve the accuracy of abnormality identification.

[0179] Specifically, as shown in Figure 8 the misjudgment of the power domain abnormal event is investigated to obtain a misjudgment event, specifically including the following steps:

[0180] S721, obtaining an optical fiber position where the power domain abnormal event exists and the voltage domain abnormal event does not exist;

[0181] In the embodiment, in the optical fiber position where the power domain abnormal event exists and the voltage domain abnormal event does not exist, there may be a misjudgment event. Therefore, the optical fiber position needs to be obtained, and the event corresponding to the optical fiber position needs to be analyzed to determine whether there is a misjudgment power domain abnormal event.

[0182] S722, extracting an OTDR voltage domain data segment of a preset length centered on the optical fiber position from the OTDR data;

[0183] In the embodiment, a local voltage domain data segment of a preset number of OTDR voltage domain data points centered on the optical fiber position is intercepted from the OTDR data as analysis data for analyzing the misjudgment event.

[0184] S723, obtaining a signal-to-noise ratio attenuation value of the OTDR voltage domain data segment, and if the signal-to-noise ratio attenuation value does not exceed a preset attenuation threshold, determining the power domain abnormal event corresponding to the optical fiber position as a misjudgment event;

[0185] In the embodiment, the signal of the optical fiber in the transmission process will appear normal attenuation phenomenon, which will not cause great adverse effects on the optical fiber transmission, so the normal attenuation phenomenon of the optical fiber is allowed to exist and does not need to be avoided by additional processing. In the OTDR event identification, the normal attenuation phenomenon may be misjudged as an abnormal event, so the normal attenuation event needs to be deleted in the power domain abnormal event to reduce subsequent invalid investigation. Understandably, if the signal-to-noise ratio attenuation value does not exceed the preset attenuation threshold, it proves that the normal attenuation event occurs at the optical fiber position, so the normal attenuation event of the optical fiber position is determined as a misjudgment event. Preferably, the preset attenuation threshold can be 5db, which can be adjusted according to the noise of the OTDR data.

[0186] S724, obtaining a jitter index of the OTDR voltage domain data segment, and if the jitter index exceeds a preset jitter threshold, determining the power domain abnormal event corresponding to the optical fiber position as a misjudgment event.

[0187] In the embodiment, the signal of the optical fiber in the transmission process will appear normal attenuation phenomenon, which will not cause great adverse effects on the optical fiber transmission, so the normal attenuation phenomenon of the optical fiber is allowed to exist and does not need to be avoided by additional processing. In the OTDR event identification, the normal attenuation phenomenon may be misjudged as an abnormal event, so the normal attenuation event needs to be deleted in the power domain abnormal event to reduce subsequent invalid investigation. Understandably, if the signal-to-noise ratio attenuation value does not exceed the preset attenuation threshold, it proves that the normal attenuation event occurs at the optical fiber position, so the normal attenuation event of the optical fiber position is determined as a misjudgment event. Preferably, the preset attenuation threshold can be 5db, which can be adjusted according to the noise of the OTDR data.

[0188] In the specific implementation process, steps S723 and S724 can be performed simultaneously, or one of them can be selected according to the actual situation, which is not limited here.

[0189] S730, deleting the misjudgment event from the power domain abnormal event to obtain an updated power domain abnormal event;

[0190] Preferably, in the identified power domain abnormal event and / or voltage domain abnormal event, when the data length of the first abnormal event is twice the length of the previous abnormal event, the phenomenon is called ghosting phenomenon, the first abnormal event is only the shadow of the previous abnormal event, and is not another abnormal event, so the first abnormal event appearing the ghosting phenomenon is also determined as a misjudgment event, which is deleted subsequently.

[0191] S740, obtaining abnormal event information in the OTDR data according to the voltage domain abnormal event and the updated power domain abnormal event;

[0192] In the embodiment, the abnormal event information at least includes power domain detection data and voltage domain detection data corresponding to the abnormal event, an event label corresponding to the abnormal event, and a probability value corresponding to the abnormal event; preferably, the abnormal event information can also be displayed by being marked in a curve graph corresponding to the OTDR data, so as to visually mark the abnormal event and improve the efficiency of subsequent troubleshooting.

[0193] S750, obtaining the OTDR event detection result according to the abnormal event information.

[0194] In the embodiment, the OTDR event detection result at least includes a plurality of abnormal event information and a visual abnormal event graph, so as to facilitate subsequent positioning, troubleshooting and repair of the abnormal event.

[0195] As shown in Figure 9 The embodiment of the application further provides an OTDR event detection system. Optionally, the system includes:

[0196] An OTDR data obtaining module 811, a data block obtaining module 812, an abnormality obtaining module 813, a clustering module 814, an original data segment obtaining module 815, a detection data obtaining module 816, and a detection result obtaining module 817. Specifically:

[0197] The OTDR data obtaining module 811 is configured to obtain OTDR data.

[0198] In the embodiment, the OTDR data obtaining module 811 can be configured to perform step S100 as shown in Figure 2 The specific description of the OTDR data obtaining module 811 can refer to the description of step S100.

[0199] The data block obtaining module 812 is configured to perform block processing on the OTDR data to obtain a plurality of data blocks; wherein each data block includes a plurality of feature data points and a plurality of verification data points.

[0200] In the embodiment, the data block obtaining module 812 can be configured to perform step S200 as shown in Figure 2 The specific description of the data block obtaining module 812 can refer to the description of step S200.

[0201] The abnormality obtaining module 813 is configured to obtain a predicted data point corresponding to the verification data point according to the feature data point of each data block, and obtain an abnormal data point according to the predicted data point and the verification data point of each data block; wherein the abnormal data point carries position information.

[0202] In the embodiment, the abnormality obtaining module 813 can be configured to perform step S300 as shown in Figure 2The step S300 is shown, and the specific description of the abnormality obtaining module 813 can refer to the description of the step S300.

[0203] The clustering module 814 is configured to obtain a plurality of abnormal data clusters from the OTDR data according to the abnormal data points and the corresponding position information.

[0204] In this embodiment, the clustering module 814 can be configured to perform Figure 2 The step S400 is shown, and the specific description of the clustering module 814 can refer to the description of the step S400.

[0205] The original data segment obtaining module 815 is configured to obtain an original data segment containing a preset data amount from the OTDR data according to the abnormal data cluster.

[0206] In this embodiment, the original data segment obtaining module 815 can be configured to perform Figure 2 The step S500 is shown, and the specific description of the original data segment obtaining module 815 can refer to the description of the step S500.

[0207] The detection data obtaining module 816 is configured to obtain corresponding detection data according to the original data segment.

[0208] In this embodiment, the detection data obtaining module 816 can be configured to perform Figure 2 The step S600 is shown, and the specific description of the detection data obtaining module 816 can refer to the description of the step S600.

[0209] The detection result obtaining module 817 is configured to input the detection data into an event detection model to perform event detection, and obtain an OTDR event detection result.

[0210] In this embodiment, the detection result obtaining module 817 can be configured to perform Figure 2 The step S700 is shown, and the specific description of the detection result obtaining module 817 can refer to the description of the step S700.

[0211] The embodiments of the present application also provide an electronic device, and the structure thereof is shown as Figure 10 The electronic device includes a memory 911, a processor 912, a communication module 913, an input / output interface 914, and the like. Optionally, the memory 911, the processor 912, the communication module 913, and the input / output interface 914 can be connected and communicated through a bus 915.

[0212] The memory 911 is configured to store one or more computer programs and transmit codes of the computer programs to the processor 912; when the one or more computer programs are executed by the processor 912, a method for detecting an OTDR event in the embodiments of the present application is implemented.

[0213] Optionally, the electronic device can be connected to a network through the communication module 913, so as to communicate with other devices such as terminals or servers through the network, and realize the interaction of data. The electronic device can be various forms of digital computers, such as desktop computers, servers, workstations, mainframe computers or other types of computers. The electronic device can also be various forms of mobile terminals, such as smart phones, tablet computers, wearable devices (such as helmets, glasses, watches, etc.) and other similar mobile terminals.

[0214] Optionally, the electronic device can connect the required input / output devices such as keyboards, display devices, etc. through the input / output interface 914, and the electronic device itself can have a display device and can also be externally connected to other display devices through the input / output interface 914. Optionally, the storage device such as hard disk, etc. can be connected through the input / output interface 914, so that the data in the electronic device can be stored in the storage device or the data in the storage device can be read, and the data in the storage device can also be stored in the memory 911. It can be understood that the input / output interface 914 can be a wired interface or a wireless interface. According to different actual application scenarios, the devices connected with the input / output interface 914 can be a component of the electronic device, or an external device connected with the electronic device when needed.

[0215] Optionally, the memory 911 can be a volatile memory and / or a non-volatile memory, the volatile memory can be a random access memory, etc., and the non-volatile memory can be a read-only memory, a programmable read-only memory, an erasable programmable read-only memory, an electrically erasable programmable read-only memory or a flash memory, etc.

[0216] Optionally, the computer programs stored in the memory 911 can be divided into one or more modules, the one or more modules are stored in the memory 911 and executed by the processor 912 to complete the method provided in the embodiments. The one or more modules can be a series of computer program instruction segments capable of completing a specific function, and the computer program instruction segments are used to describe the execution process of the computer program in the electronic device.

[0217] Optionally, the processor 912 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 912 include, but are not limited to, a central processing unit, a graphics processing unit, a digital signal processor, various special-purpose artificial intelligence computing chips, various processors running machine learning model algorithms, and can also be any appropriate controller, microcontroller, processor, etc. The processor 912 executes various methods and processes of the embodiments, for example, an OTDR event detection method of the embodiments.

[0218] Optionally, the bus 915 can include a path for transmitting information. According to different functions, the bus 915 can be divided into an address bus, a data bus, a control bus, etc.

[0219] In an optional implementation, the embodiments of the present application also provide a computer storage medium having a computer program stored thereon, and the computer program enables a computer to execute the method of the method embodiments when executed by the computer. Part or all of the computer program can be loaded and / or installed on the memory 911 of the electronic device. When the computer program is executed by the processor 912, one or more steps of an OTDR event detection method of the embodiments of the present application can be executed.

[0220] Optionally, the computer-readable storage medium can be a random access memory, a read-only memory, a programmable read-only memory, an erasable programmable read-only memory, an electrically erasable programmable read-only memory, etc.

[0221] Obviously, the above-described embodiments of the present application are only examples for clearly illustrating the technical solutions of the present application, and are not intended to limit the specific embodiments of the present application. Any modification, equivalent replacement, and improvement made within the spirit and principle of the claims of the present application shall be included in the protection scope of the claims of the present application.

Claims

1. A method of OTDR event detection, characterized by, The method comprises: acquiring OTDR data; performing block processing on the OTDR data to obtain a plurality of data blocks; wherein each data block comprises a plurality of feature data points and a plurality of verification data points; acquiring, according to the feature data points of each data block, a predicted data point corresponding to the verification data point, and acquiring, according to the predicted data point and the verification data point of each data block, an abnormal data point; wherein the abnormal data point carries position information; acquiring, according to the abnormal data point and the corresponding position information, a plurality of abnormal data clusters from the OTDR data; acquiring, according to the abnormal data cluster, a raw data segment containing a preset data amount from the OTDR data; acquiring corresponding detection data according to the raw data segment; inputting the detection data into an event detection model for event detection to obtain an OTDR event detection result.

2. The method of claim 1, wherein, The block processing on the OTDR data to obtain a plurality of data blocks; wherein each data block comprises a plurality of feature data points and a plurality of verification data points, comprises: performing block processing on the OTDR data according to a preset total number of data blocks and a preset moving step number to obtain a plurality of data blocks; wherein the number of data points in each data block is the total number of data blocks; the first data block takes the first data point of the OTDR data as the division starting point, and the division starting point of other data blocks is separated from the division starting point of the previous data block by the moving step number; dividing the data blocks according to a preset ratio to obtain a plurality of feature data points and a plurality of verification data points corresponding to the data blocks.

3. The method of claim 1, wherein, The acquisition, according to the feature data points of each data block, of a predicted data point corresponding to the verification data point, and the acquisition, according to the predicted data point and the verification data point of each data block, of an abnormal data point, comprises: performing data prediction according to the feature data points to obtain a plurality of predicted data points, one predicted data point corresponding to one verification data point; obtaining a difference value by subtracting the predicted data point from the corresponding verification data point; obtaining a maximum difference value in the difference value; determining an abnormal data point according to the difference value corresponding to the verification data point and the maximum difference value.

4. The method of claim 3, wherein, The determination of an abnormal data point according to the difference value corresponding to the verification data point and the maximum difference value, comprises: presetting a first difference threshold and a second difference threshold; the second difference threshold is greater than the first difference threshold; determining whether the difference value corresponding to the verification data point exceeds the first difference threshold and whether the maximum difference value exceeds the second difference threshold, and if so, regarding the verification data point as an abnormal data point.

5. The method of claim 1, wherein, The acquisition, according to the abnormal data point and the corresponding position information, of a plurality of abnormal data clusters from the OTDR data, comprises: acquiring, according to the position information of two adjacent abnormal data points, two OTDR data points with the same position information in the OTDR data; acquiring the number of OTDR data points between the two OTDR data points; If the number of OTDR data points is less than or equal to a preset first clustering threshold, the corresponding two OTDR data points and the OTDR data points between the corresponding two OTDR data points are taken as an abnormal data cluster.

6. The method according to any one of claims 1 to 5, characterized in that, The method comprises the following steps: acquiring the number of data points of the abnormal data cluster; if the number of data points of the abnormal data cluster does not exceed a preset second clustering threshold, the first OTDR data point of the abnormal data cluster is taken as a division node; if the number of data points of the abnormal data cluster exceeds the second clustering threshold but does not exceed a preset third clustering threshold, a plurality of division nodes of the abnormal data cluster are acquired at a preset data point interval; if the number of data points of the abnormal data cluster exceeds the third clustering threshold, a plurality of OTDR data points of a preset third data point number are extracted from the end of the corresponding abnormal data cluster to update the corresponding abnormal data cluster, and a plurality of division nodes are acquired in the updated abnormal data cluster at the data point interval; acquiring a raw data segment containing a preset data amount from the OTDR data corresponding to the division node according to the position information of the division node.

7. The method according to any one of claims 1 to 5, characterized in that, The detection data includes power domain detection data and voltage domain detection data. The method comprises the following steps: inputting the detection data into an event detection model for event detection to obtain an OTDR event detection result, comprising: inputting the power domain detection data and the voltage domain detection data into an event detection model for event recognition to obtain a plurality of power domain abnormal events and a plurality of voltage domain abnormal events; performing misjudgment investigation on the power domain abnormal events to obtain misjudgment events; deleting the misjudgment events from the power domain abnormal events to obtain updated power domain abnormal events; obtaining abnormal event information in the OTDR data according to the voltage domain abnormal events and the updated power domain abnormal events; 8. The method of claim 7, wherein, obtaining the OTDR event detection result according to the abnormal event information. The method comprises the following steps: obtaining the OTDR event detection result according to the abnormal event information. The method comprises the following steps:

9. An OTDR event detection system characterized by, obtaining an optical fiber position where the power domain abnormal event exists and the voltage domain abnormal event does not exist; extracting an OTDR voltage domain data segment of a preset length centered on the optical fiber position from the OTDR data; obtaining a signal-to-noise ratio decay value of the OTDR voltage domain data segment, if the signal-to-noise ratio decay value does not exceed a preset decay threshold, the power domain abnormal event corresponding to the optical fiber position is determined as a misjudgment event; and / or, obtaining a jitter index of the OTDR voltage domain data segment, if the jitter index exceeds a preset jitter threshold, the power domain abnormal event corresponding to the optical fiber position is determined as a misjudgment event. The system comprises: an OTDR data acquisition module for acquiring OTDR data; a data block acquisition module for block processing the OTDR data to obtain a plurality of data blocks; wherein the data block comprises a plurality of feature data points and a plurality of verification data points; anomaly obtaining module, configured to obtain a predicted data point corresponding to the verification data point according to a feature data point of each data block, and obtain an anomaly data point according to the predicted data point and the verification data point of each data block, wherein the anomaly data point carries position information; clustering module, configured to obtain a plurality of anomaly data clusters from the OTDR data according to the anomaly data points and the corresponding position information; original data segment obtaining module, configured to obtain an original data segment containing a preset data amount from the OTDR data according to the anomaly data cluster; detection data obtaining module, configured to obtain corresponding detection data according to the original data segment; detection result obtaining module, configured to input the detection data into an event detection model to perform event detection, and obtain an OTDR event detection result.

10. An electronic device, comprising: comprise: a memory, configured to store one or more computer programs; a processor, when the one or more computer programs are executed by the processor, implements an OTDR event detection method according to any one of claims 1-8.

Images