OPGW optical cable fiber core strain classification method and system based on supervised machine learning

Abstract

The invention discloses an OPGW optical cable fiber core strain classification method and system based on supervised machine learning, and relates to the technical field of optical cable detection. The method comprises the following steps: firstly, carrying out statistical analysis on strain test data of an OPGW optical cable in a typical area, and providing four typical strain forms of the optical cable: a zero strain form, a catenary form, a standing wave form and a local large strain form; the measured OPGW optical cable fiber core strain is classified according to forms, labels are given, and a data set is provided for subsequent OPGW optical cable classification; training the CNN model by using the data set; the trained CNN model is applied to new optical cable fiber core strain data for strain classification, and the classification accuracy is 99.2%; furthermore, the new optical cable fiber core strain generation reason is judged according to the strain form, so that the operation and maintenance personnel can maintain the fault optical cable section. According to the method, the OPGW optical cable faults can be effectively detected and classified quickly and accurately, and an important reference is provided for strain data analysis in subsequent engineering application.

Description

Technical Field

[0001] The present invention relates to the technical field of optical cable detection, and in particular to an OPGW optical cable core strain classification method and system based on supervised machine learning. Background Art

[0002] OPGW optical cables need to withstand tension and gravity during operation. Long-term stress will cause the cables to stretch. At the same time, affected by thermal expansion and contraction caused by temperature and external environments such as wind vibration, icing, and lightning strikes, the optical cables are also susceptible to stress and inelastic deformation, which affects the service life of the optical cables. When the OPGW optical cable is stretched, the fiber core inside its optical unit will stretch along with the optical cable, which is manifested as increased strain. Long-term strain of the optical cable core will affect the performance and service life of the optical cable. In severe cases, there will be risks of fiber breakage, strand breakage, or tower collapse. By monitoring the strain distribution of the OPGW spare fiber core using distributed Brillouin optical fiber sensing technology, strain monitoring of the entire OPGW optical cable can be achieved.

[0003] Machine learning is an algorithm based on statistical theory, which aims to derive rules from given data and use these rules to predict unknown data. Supervised learning requires a pre-labeled data set and establishes a mapping relationship between input and output. Supervised learning can be used for regression analysis and classification tasks, where regression analysis is used for continuous value prediction and classification tasks are used for discrete label prediction. In OPGW optical cable monitoring, not only a large amount of monitoring data provided by distributed fiber optic sensing technology is needed to ensure the safety of optical cable operation, but also machine learning is needed to process these monitoring data and deeply explore the value of the data.

[0004] At present, the machine learning technologies used in OPGW optical cable monitoring include: Brillouin spectrum signal demodulation technology, temperature-strain decoupling technology, and OPGW optical cable fault warning technology. Brillouin spectrum signal demodulation technology uses machine learning algorithms to replace traditional Brillouin spectrum fitting algorithms to solve Brillouin frequency shift or temperature. Generally, the same optical fiber is used to establish the corresponding relationship between its Brillouin spectrum and Brillouin frequency shift / temperature, and then the new Brillouin spectrum predicts the optical fiber Brillouin frequency shift / temperature; this model needs to be retrained after replacing the optical fiber, is not universal, and is difficult to use in actual projects. The essence of temperature-strain decoupling technology is to use the characteristics of the LEAF optical fiber double-peak spectrum to achieve temperature and strain demodulation, and it cannot be used in OPGW optical cables. Machine learning algorithms have been initially applied in warnings of events such as temperature and icing, but due to the complex actual test environment, more data is needed to optimize the model to make the prediction effect more accurate.

[0005] In summary, although machine learning algorithms have been preliminarily applied in OPGW optical cable monitoring, they are subject to certain limitations in actual engineering applications and need further research. Summary of the invention

[0006] To this end, the present invention proposes an OPGW optical cable core strain classification method and system based on supervised machine learning, in an effort to solve or alleviate one or more of the above problems.

[0007] According to one aspect of the present invention, a method for classifying OPGW optical cable core strain based on supervised machine learning is proposed, the method comprising:

[0008] Use BOTDA system or BOTDR system to collect Brillouin frequency shift of OPGW optical cable;

[0009] Acquire a core strain data sequence of each optical cable segment according to the Brillouin frequency shift demodulation;

[0010] Analyze and process the core strain data sequence to obtain the core strain morphology type corresponding to each optical cable segment; wherein the core strain morphology type includes zero strain morphology, catenary morphology, standing wave morphology, and local large strain morphology;

[0011] The fiber core strain morphology type is used as a classification label and input into a deep learning model together with the fiber core strain data sequence for training to obtain a trained classification model based on deep learning;

[0012] The optical cable core strain data sequence corresponding to the OPGW optical cable to be tested, collected by the BOTDA system or the BOTDR system, is input into the trained classification model based on deep learning to obtain the core strain morphology type corresponding to each optical cable segment in the OPGW optical cable to be tested.

[0013] Further, the method of obtaining the core strain data sequence of each optical cable segment according to the Brillouin frequency shift calculation includes: obtaining multiple optical cable segments of different lengths, including: assuming that the core length of the optical cable in the tension section is L, the heights of the tension towers on both sides are h1 and h2, and the down conductor is reserved for welding, then the core length of the optical cable in the tension section after removing the down conductor is L fiber = L-h1-h2-2h; Determine the number of connected towers n and the spacing l between two adjacent towers in the tension section according to the tower list i , i=1,2,…,n-1, calculate the cumulative span in the tension section Calculate the ratio of the optical cable core length to the cumulative span length k = L fiber / L tower , the core length of the optical cable is L i =k×l i , thereby obtaining multiple optical cable segments of different lengths; based on the different lengths of the divided optical cable segments, the Brillouin frequency shift is demodulated to obtain the core strain data sequences of the multiple optical cable segments.

[0014] Further, the analyzing and processing the fiber core strain data sequence to obtain the fiber core strain morphology type corresponding to each optical cable segment includes:

[0015] If each strain value in the fiber core strain data sequence is less than the preset threshold, it indicates that the optical fiber is not stressed, and the fiber core strain morphology type corresponding to the optical cable segment is zero strain morphology;

[0016] If the strain distribution corresponding to the core strain data sequence conforms to the catenary equation, and the maximum strain value exceeds the preset threshold, the core strain morphology type corresponding to the optical cable segment is the catenary morphology;

[0017] If the strain distribution corresponding to the core strain data sequence is in the form of a periodic standing wave or conforms to the superposition equation of the catenary form and the standing wave form, and the maximum strain value exceeds the preset threshold, then the core strain form type corresponding to the optical cable section is the standing wave form;

[0018] If a local strain value that is short-lived and far higher than a preset threshold corresponding to a zero strain morphology appears in the strain distribution corresponding to the core strain data sequence, the core strain morphology type corresponding to the optical cable segment is a local large strain.

[0019] Furthermore, the catenary equation is:

[0020]

[0021] Where y represents strain, x represents the position of the optical cable, It represents the angle between the slant span and the x-axis, l represents the span between the two towers; σ0 represents the horizontal stress at the lowest point of the sag; γ represents the specific load of the optical cable.

[0022] Furthermore, the superposition equation of the catenary shape and the standing wave shape is:

[0023]

[0024] Where y represents strain, x represents the position of the optical cable, represents the angle between the slant span and the x-axis, l represents the span between the two towers; σ0 represents the horizontal stress at the lowest point of the sag; γ represents the specific load of the optical cable; λ represents the standing wave wavelength; and A represents the amplitude of the standing wave signal.

[0025] Furthermore, after the fiber core strain data sequence of each optical cable segment is acquired, it is preprocessed, and the preprocessing includes: performing a truncation or zero-filling operation on the fiber core strain data sequence of each optical cable segment.

[0026] Furthermore, the deep learning model is a convolutional neural network; the core strain morphology type is used as a classification label and input into the deep learning model together with the core strain data sequence for training, including: the core strain data sequence passes through 3 convolution modules respectively, each convolution module includes a convolution layer, a ReLU activation function and a maximum pooling layer, that is, passes through 32 cores, 64 cores, 128 cores and corresponding 3 maximum pooling layers; then enters the fully connected layer, the fully connected layer includes two hidden layers, and the data output by the maximum pooling layer is flattened into a one-dimensional sequence; the final output layer contains 4 nodes, corresponding to 4 types of core strain morphology.

[0027] Furthermore, after obtaining the fiber core strain morphology type corresponding to each optical cable segment in the OPGW optical cable to be tested, the following analysis and processing are performed on it:

[0028] If the fiber core strain morphology type corresponding to a certain optical cable segment is a catenary shape, then the strain value in the fiber core strain data sequence is used to determine whether the optical cable segment is iced in combination with the temperature;

[0029] If the core strain morphology type corresponding to a certain optical cable segment is a standing wave morphology, then based on the strain value in the core strain data sequence and combined with the field observation results, it is further determined whether the optical cable segment has loose strands and twisted wire breaks;

[0030] If the strain morphology type of the fiber core corresponding to a certain optical cable segment is a local large strain morphology, then based on the on-site observation results, it is further determined whether the strands at the clamping position of the tower top clamp corresponding to the optical cable segment are deformed.

[0031] According to another aspect of the present invention, a OPGW optical cable core strain classification system based on supervised machine learning is proposed, the system comprising:

[0032] A signal acquisition module configured to acquire the Brillouin frequency shift of the OPGW optical cable using a BOTDA system or a BOTDR system;

[0033] A signal processing module, configured to obtain a core strain data sequence of each optical cable segment according to the Brillouin frequency shift demodulation; analyze and process the core strain data sequence to obtain a core strain morphology type corresponding to each optical cable segment; wherein the core strain morphology type includes a zero strain morphology, a catenary morphology, a standing wave morphology, and a local large strain morphology;

[0034] A model training module is configured to input the core strain morphology type as a classification label and the core strain data sequence into a deep learning model for training, thereby obtaining a trained classification model based on deep learning;

[0035] The strain morphology classification module is configured to input the optical cable core strain data sequence corresponding to the OPGW optical cable to be tested, which is collected by the BOTDA system or the BOTDR system, into the trained deep learning-based classification model to obtain the core strain morphology type corresponding to each optical cable segment in the OPGW optical cable to be tested.

[0036] Further, the signal processing module analyzes and processes the core strain data sequence to obtain the core strain morphology type corresponding to each optical cable segment, including: if each strain value in the core strain data sequence is less than a preset threshold, it indicates that the optical fiber is not stressed, and the core strain morphology type corresponding to the optical cable segment is a zero strain morphology;

[0037] If the strain distribution corresponding to the core strain data sequence conforms to the catenary equation, and the maximum strain value exceeds the preset threshold, the core strain morphology type corresponding to the optical cable segment is the catenary morphology;

[0038] If the strain distribution corresponding to the core strain data sequence is in the form of a periodic standing wave or conforms to the superposition equation of the catenary form and the standing wave form, and the maximum strain value exceeds the preset threshold, then the core strain form type corresponding to the optical cable section is the standing wave form;

[0039] If a local strain value that is short-lived and far higher than a preset threshold corresponding to a zero strain morphology appears in the strain distribution corresponding to the core strain data sequence, the core strain morphology type corresponding to the optical cable segment is a local large strain.

[0040] The beneficial technical effects of the present invention are:

[0041] The present invention proposes a method and system for classifying OPGW optical cable core strain based on supervised machine learning. The present invention first statistically analyzes the strain test data of OPGW optical cables in typical areas, and proposes four typical strain forms of optical cables: zero strain form, catenary form, standing wave form and local large strain form. By performing stress analysis on OPGW optical cables, the correctness of the classification of these four stress forms is verified, and the causes of optical cable failures corresponding to different strain forms are analyzed. The typical core strain form analysis of OPGW optical cables classifies the measured OPGW optical cable core strain according to form and gives labels, providing a data set for subsequent OPGW optical cable classification; then, the CNN model is trained using the data set, and the model strain classification accuracy is 93.01%. The trained model is applied to a new optical cable core strain data for strain classification, and the accuracy is 99.2%. The strain classification results show that the CNN model can effectively classify the strain form of OPGW optical cables. Furthermore, the cause of the new optical cable core strain is judged according to the strain form, which helps the operation and maintenance personnel to repair the faulty optical cable section. The present invention can effectively detect and classify OPGW optical cable faults quickly and accurately, and provide an important reference for strain data analysis in subsequent engineering applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood by reading the detailed description below with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, in which:

[0043] Figure 1 It is a flow chart of an OPGW optical cable core strain classification method based on supervised machine learning described in an embodiment of the present invention.

[0044] Figure 2 It is a schematic diagram of the structure of a long-distance BOTDA / BOTDR fusion system in an embodiment of the present invention.

[0045] Figure 3 It is a schematic diagram of the OPGW optical cable installation method in an embodiment of the present invention.

[0046] Figure 4 It is a structural diagram of the OPGW optical cable in an embodiment of the present invention.

[0047] Figure 5 It is the catenary force diagram of the OPGW optical cable in the embodiment of the present invention.

[0048] Figure 6 It is a diagram of the strain simulation results of the catenary morphology of the OPGW optical cable in an embodiment of the present invention.

[0049] Figure 7Schematic diagram of standing wave morphology strain of OPGW optical cable in an embodiment of the present invention.

[0050] Figure 8 It is a diagram of the simulation results of the standing wave morphology strain of the OPGW optical cable in an embodiment of the present invention.

[0051] Fig. 9 It is a force diagram of the rigid OPGW optical cable in an embodiment of the present invention.

[0052] Fig.10 This is a simulation result diagram of the local large strain morphology of the OPGW optical cable in an embodiment of the present invention.

[0053] Fig.11 1 is a typical strain example diagram of the OPGW optical cable in the embodiment of the present invention; wherein a) corresponds to the zero strain form; b) corresponds to the catenary form; c) corresponds to the standing wave form; d) corresponds to the local large strain form.

[0054] Fig.12 It is a schematic diagram of the CNN structure in an embodiment of the present invention. DETAILED DESCRIPTION

[0055] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present invention, and are not intended to limit the scope of the present invention in any way. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0056] Those skilled in the art know that the embodiments of the present invention can be implemented as a system, device, apparatus, method or computer program product. Therefore, the present disclosure can be specifically implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software. In the present invention, it is to be understood that any number of elements in the drawings is for illustration and not for limitation, and any naming is only for distinction and does not have any limiting meaning.

[0057] The embodiment of the present invention proposes an OPGW optical cable core strain classification method based on supervised machine learning, such as Figure 1 As shown, the method includes:

[0058] S1. Use BOTDA system or BOTDR system to collect Brillouin frequency shift of OPGW optical cable;

[0059] S2. Acquire a core strain data sequence of each optical cable segment according to the Brillouin frequency shift demodulation;

[0060] S3, analyzing and processing the core strain data sequence to obtain the core strain morphology type corresponding to each optical cable segment; wherein the core strain morphology type includes zero strain morphology, catenary morphology, standing wave morphology, and local large strain morphology;

[0061] S4, taking the core strain morphology type as a classification label and inputting it together with the core strain data sequence into a deep learning model for training, thereby obtaining a trained classification model based on deep learning;

[0062] S5. Inputting the optical cable core strain data sequence corresponding to the OPGW optical cable to be tested collected by the BOTDA system or the BOTDR system into the trained classification model based on deep learning, and obtaining the core strain morphology type corresponding to each optical cable segment in the OPGW optical cable to be tested;

[0063] After obtaining the core strain morphology type corresponding to each optical cable segment in the OPGW optical cable to be tested, S6 performs the following analysis and processing: if the core strain morphology type corresponding to a certain optical cable segment is a catenary morphology, then according to the strain value in the core strain data sequence and in combination with the temperature, it is determined whether the optical cable segment is iced; if the core strain morphology type corresponding to a certain optical cable segment is a standing wave morphology, then according to the strain value in the core strain data sequence and in combination with the field observation results, it is further determined whether the optical cable segment has loose strands and twisted wire breakage; if the core strain morphology type corresponding to a certain optical cable segment is a local large strain morphology, then according to the field observation results, it is further determined whether the twisted wire at the clamping position of the tower top wire clamp corresponding to the optical cable segment is deformed.

[0064] The method starts from S1. In S1, a BOTDA system or a BOTDR system is used to collect the Brillouin frequency shift of the OPGW optical cable.

[0065] According to an embodiment of the present invention, a long-distance BOTDA system based on Raman amplification, a long-distance BOTDR system based on Raman amplification, or a long-distance BOTDR sensing system based on remote pump amplification in the prior art is used to collect 12 months of Brillouin frequency shift data of a 500kV OPGW optical cable across the country or in a certain city.

[0066] In this embodiment, preferably, a long-distance BOTDA / BOTDR fusion system prototype is built to support two test modes: single-ended measurement and double-ended loop measurement. The system solution is as follows: Figure 2As shown. The upper branch is used to generate pump light. Its structure is the same as that of BOTDA and BOTDR systems. 50% of the continuous light output by the laser is modulated into pulse light, which is then scrambled and amplified to generate pump light and injected into the optical fiber to be tested. Brillouin scattered light enters EDFA2 through port 3 of circulator 1. Since the Brillouin scattering signal intensity of BOTDA and BOTDR is quite different, it is necessary to adjust the amplification factor through EDFA2 to ensure that the optical signal intensity received by the detector is high and will not be saturated. The lower branch uses EOM to modulate 50% of the continuous light output by the laser into double-sideband detection light. After coupler 3, 10% is separated for wavelength calibration, and the remaining 90% of the light is used as the detection light of BOTDA or the local reference light of BOTDR. Its output power is controlled by VOA. When working in BOTDA mode, the output power of the detection light is about 1mW. When working in BOTDR mode, the output power of the local reference light is about 200μW. Too high power will cause detector saturation, resulting in incorrect measurement results in BOTDR mode.

[0067] The fusion system prototype has three working modes, and the working mode switching is achieved by controlling three optical switches (OS). In BOTDA mode, OS1 is connected to 2 channels, and the PD detector of BOTDA is used to detect the signal. OS2 is connected to 2 channels, and the double-sideband detection light enters the optical fiber to be tested from the probe end of the fusion system to achieve double-ended measurement. OS3 is connected to 1 channel, so that the stimulated Brillouin scattering signal enters the detector after passing through the circulator and filter. In BOTDR mode, OS3 is connected to 1 channel, so that the spontaneous Brillouin scattering signal enters the four-port circulator after passing through the circulator and filter. OS1 is connected to 1 channel, and OS2 is connected to 1 channel, so that the spontaneous Brillouin scattering light and the local reference light enter the four-port coupler to coherently, and are detected by a balanced detector to achieve single-ended measurement. Affected by the external ambient temperature, the laser wavelength may drift. When the laser and filter wavelengths do not match, the system cannot work normally, and the laser wavelength needs to be calibrated regularly. In the laser wavelength calibration mode, OS3 is connected to 2 channels, and 10% of the double-sideband detection light enters the filter through circulator 2. Since the FBG in the filter only reflects light of a specific wavelength and outputs it from the circulator 3 port, the remaining invalid signals will pass through the filter. The laser wavelength calibration can be achieved by monitoring the intensity of the filter transmitted light signal.

[0068] When calibrating the laser wavelength, first change the laser control voltage to scan the laser wavelength over a wide range so that both of its sidebands can cover the filter operating wavelength. When the wavelength of the sideband is aligned with the filter wavelength, the sideband signal is reflected by the filter, and the intensity of the transmitted light signal is reduced. Since the detection light has two sidebands, the filter transmission signal has two lowest points during the laser wavelength calibration process. The laser wavelength calibration can be completed by finding the voltage value corresponding to the lowest point through the peak-finding algorithm. When calibrating the laser wavelength, the control voltage is changed at equal intervals, and the laser output wavelength gradually increases with the increase of the control voltage. Therefore, the second lowest point with a larger wavelength is selected, and the corresponding transmitted light is Stokes light. The Brillouin spectrum obtained by BOTDA and BOTDR tests is the gain spectrum. This can achieve automatic calibration of the laser wavelength of the fusion system.

[0069] The fusion system prototype is composed of modular hardware, including power supply, mainboard, acquisition card, laser, microwave source, pulse source, EDFA, detector and optical passive module. The optical passive module integrates multiple optical devices such as AOM, EOM, VOA, isolator, coupler, circulator, filter and optical switch, so that the structure of the fusion system is optimized. The prototype adopts a layered structure design, with the lower layer being the optical subsystem, including Figure 2 All optical and electrical modules of the optical path diagram of the fusion system. The upper layer is the data acquisition and processing subsystem, including the data acquisition card, mainboard and power supply, which is used to demodulate the Brillouin scattering signal into Brillouin frequency shift, temperature or strain. Serial communication is used between the upper and lower layers. The mainboard sends serial port commands to control the hardware modules in the optical subsystem, complete the pulse setting, sweep frequency setting, laser wavelength setting, EDFA working current setting and optical switch channel setting and other hardware module state control, and finally obtain the Brillouin frequency shift. Since the long-distance BOTDA and BOTDR systems are built using modules, the fusion system prototype can achieve the same performance indicators as the BOTDA and BOTDR systems.

[0070] Then, S2 is executed, in which a core strain data sequence of each optical cable segment is acquired according to the Brillouin frequency shift demodulation.

[0071] According to an embodiment of the present invention, for an OPGW optical cable core with strain, the strain is segmented by span, and the strain of the optical cable core is analyzed in units of span. The length of the optical fiber of the down-lead at both ends is removed, and the strain is segmented by span according to the span of the tower in that section. The algorithm is as follows:

[0072] First, assuming that the length of the optical cable core in the tension section is L, the heights of the tension towers on both sides are h1 and h2, and the down conductor is reserved for h (h = 10m) for welding, then the length of the optical cable core in the tension section after removing the down conductor is L fiber =L-h1-h2-2h;

[0073] Then, query the tower list to determine the number of connected towers n and the spacing l between two adjacent towers in the tension section. i , where i = 1, 2, ..., n-1, and calculate the cumulative span in the tension section

[0074] Then, calculate the ratio of the optical cable core length to the cumulative span length k = L fiber / L tower , the core length of the optical cable is L i =k×l i , where i=1,2,…,n-1.

[0075] The core strain measurement results are segmented by tension section using the above algorithm. After obtaining multiple optical cable segments of different lengths, the Brillouin frequency shift is demodulated based on the different lengths of the divided optical cable segments to obtain the core strain data sequences of the multiple optical cable segments.

[0076] Specifically, since the Brillouin frequency shift is affected by both temperature and strain, temperature compensation is required when performing strain demodulation on OPGW optical cables to remove the effects of temperature. The traditional strain demodulation solution is to lay a temperature optical cable next to the strain optical cable, where the strain optical cable is affected by both temperature and strain, while the temperature optical cable is only affected by temperature. The temperature change is calculated through the temperature compensation optical fiber to accurately demodulate the strain. OPGW optical cables use existing optical fibers for strain sensing and cannot add temperature compensation optical fibers, so the traditional strain demodulation solution is not suitable for OPGW optical cable strain monitoring.

[0077] An OPGW optical cable line is formed by welding multiple optical cable segments together. There are dozens of meters of down conductors between adjacent optical cable segments, which are used to connect two sections of optical cable. Figure 3 As shown. Since the down lead is between two strain clamps and fixed on the cable drum, the optical fiber in the down lead will not produce strain, and the Brillouin frequency shift is only affected by temperature. The optical fiber at the down lead can be used as a temperature reference point to perform strain demodulation on the optical cable in the strain section. Since a strain section and the down leads on both sides are a whole section of optical cable, the Brillouin frequency shift of the optical fiber is the same under initial conditions. After hanging on the tower, the down leads on both sides are only affected by temperature, and the strain section is affected by both temperature and strain. Assuming that the Brillouin frequency shift and temperature of the left down lead are v0 and T0, and the Brillouin frequency shift and temperature of the right down lead are v1 and T1, and the temperature distribution of the entire strain section optical cable is uniform, then the core temperature of the optical cable in the strain section is T=(T0+T1) / 2, and the core strain of the optical cable in the middle section is in is the cable core gauge factor, and v represents the Brillouin frequency shift.

[0078] Then execute S3, in which the core strain data sequence is analyzed and processed to obtain the core strain morphology type corresponding to each optical cable segment; wherein the core strain morphology type includes zero strain morphology, catenary morphology, standing wave morphology, and local large strain morphology. Specifically, if each strain value in the core strain data sequence is less than the preset threshold, it indicates that the optical fiber is not stressed, and the core strain morphology type corresponding to the optical cable segment is zero strain morphology; if the strain distribution corresponding to the core strain data sequence conforms to the catenary equation, and the maximum strain value exceeds the preset threshold, the core strain morphology type corresponding to the optical cable segment is catenary morphology; if the strain distribution corresponding to the core strain data sequence is a periodic standing wave morphology or conforms to the superposition equation of the catenary morphology and the standing wave morphology, and the maximum strain value exceeds the preset threshold, the core strain morphology type corresponding to the optical cable segment is standing wave morphology; if the strain distribution corresponding to the core strain data sequence is a local and short-lasting strain value that is far higher than the preset threshold corresponding to the zero strain morphology, the core strain morphology type corresponding to the optical cable segment is local large strain.

[0079] According to the embodiment of the present invention, the strain morphology of the optical cable in a typical area is statistically analyzed in combination with the natural environment and meteorological conditions. According to morphology, the strain can be divided into four categories: zero strain morphology, catenary morphology, standing wave morphology and local large strain morphology. Since each span of the OPGW optical cable is relatively independent, the strain of the long-distance OPGW optical cable can be simulated according to the span. And for the four typical strain morphologies, a comparative analysis is conducted from the perspectives of theoretical simulation and actual test results to verify the accuracy of the actual test data.

[0080] 1) Zero strain morphology analysis

[0081] OPGW optical cables are divided into two parts according to their functions: an optical unit and a ground wire unit. The optical unit consists of a protective tube (aluminum tube or stainless steel tube) and an optical fiber. Aluminum tube protective tubes were commonly used in early OPGW optical cables, but have now been eliminated. Newly built OPGW optical cables all use stainless steel tube structures. The ground wire unit consists of two metal monofilaments, aluminum wire and aluminum-clad steel wire. The aluminum wire acts as a conductor, and the aluminum-clad steel wire acts as a support. The optical unit and the metal monofilaments are twisted together to form an OPGW optical cable. Currently, the commonly used OPGW optical cables in power grid systems are central tube structures and layer-twisted structures. The main difference between the two is whether the optical unit is a layer-twisted structure. Their structures are as follows: Figure 4 shown.

[0082] Since the allowable tensile deformation of the optical fiber is much smaller than that of the metal monofilament in the OPGW optical cable, when the optical cable is stretched, the internal fiber core of the OPGW optical cable is also stretched synchronously. In order to control the optical fiber to not produce additional loss due to the force when the OPGW is subjected to a certain range of tension, the optical fiber in the optical unit of the central tube type and layer-twisted OPGW optical cable has excess length (primary excess length), and there is grease inside, and the internal fiber core is in a free state, which can ensure that the optical fiber is not stressed under instantaneous impact and short-term large load conditions, reducing the fiber loss. At the same time, the fiber core of the layer-twisted structure optical cable has a larger excess length (secondary excess length). In recent years, the OPGW optical cables put into operation use more stainless steel layer-twisted structures. Therefore, the present invention uses the stainless steel layer-twisted structure as an example for simulation to analyze the stress of the OPGW optical cable and the fiber core.

[0083] As shown in Table 1, during the construction of OPGW optical cable, a tension machine is used at one end to release the cable from the cable drum, and a traction machine is used at the other end to stretch the cable. The traction machine and the pay-off machine work together to spread the cable. During the installation of the optical cable, the sag observation method is usually used to tighten the cable. When the sag design requirements are met, the two ends are fixed with hardware. During the entire installation process, the pulley and hardware stretch to produce a total elongation of about 0.1%. If the optical cable is pulled too tight during the tightening process and the sag is too small, the initial tension of the optical cable will be too large, consuming a lot of fiber excess length, and even showing strain. Its 30-year expected total elongation (strain) is the sum of the total expected elongation generated during the installation process and the total expected elongation generated during the operation process, which is about 0.58% to 0.63%. In order to ensure that the internal fiber core of the OPGW optical cable will not be subjected to force and produce additional attenuation during the installation and operation process, the optical cable manufacturer generally leaves a certain amount of excess length in the internal fiber core to offset the elongation of the OPGW optical cable. The excess length is generally 0.6% to 0.7%. Since the expected elongation of the optical cable during installation is 0.1%, which is less than the excess length of the optical fiber, the internal core of newly built OPGW optical cables or lines with fewer years of operation will not be strained.

[0084] Table 1 Expected elongation of OPGW optical cable [1]

[0085]

[0086] When the excess length of the OPGW cable core is exhausted, the core inside the optical unit will stretch synchronously with the cable, which is manifested as strain on the core. Huang Junhua et al. obtained the expected life of the optical fiber under different strains based on the relationship between optical fiber life and strain. [1], as shown in Table 2. In actual tests, it was found that when the strain was 0.1%, there was also a case of optical cable failure. Some thresholds in Table 2 were modified, and the optical fiber strain was divided into the following four levels according to the safety status of the optical cable: when the strain was less than 0.05%, the optical fiber was not stressed; when the strain was between 0.05% and 0.25%, the optical fiber was in a safe range and the probability of failure was low; when the strain was between 0.25% and 0.35%, the probability of optical fiber failure was high and needed special attention; when the strain was greater than 0.35%, the optical cable should be repaired or replaced in time to prevent the failure from affecting the line.

[0087] Table 2 Relationship between strain on optical fiber and optical fiber life [1]

[0088]

[0089] Test the strain of the OPGW optical cable core of a newly built flexible DC transmission line to test whether the optical cable core generates strain during the installation process, and assist in the construction acceptance of the newly built OPGW optical cable. Use the fusion system prototype to test 6 cores of the OPGW optical cable, including the 1st, 7th, 13th, 19th, 25th and 31st cores out of the 36 cores.

[0090] The Brillouin frequency shift of the newly built OPGW optical cable was tested using a fusion system prototype. There were Brillouin frequency shift jump points at the fusion joints of each section, and the Brillouin frequency shift in each section of the optical cable was relatively flat, indicating that the internal fiber core was not subjected to obvious strain during the installation of the optical cable. As shown in Table 1, only about 0.1% of the excess length was released during the installation of the new line. Due to the existence of the excess length, the internal optical fiber was stress-free in the early stage of the optical cable. When the cable body elongated more than the excess length of the optical fiber, the optical fiber showed stress. The fusion system prototype was used to test the strain of the OPGW optical cable core during the construction stage as a reference for construction acceptance to prevent excessive elongation of the optical cable caused by excessive initial tension during the installation process, and to ensure the quality of line construction.

[0091] 2) Catenary shape analysis

[0092] When OPGW optical cables are installed on two pole towers, sag will occur due to the stretching of the hardware at both ends and the gravity of the optical cable itself. Since the length of OPGW optical cables is much longer than their diameter and they are made of multiple strands of fine metal wires, the rigidity of the optical cables has little effect on the curve shape of their suspension space. They can be assumed to be ropes without rigidity, and the load acting on the OPGW optical cables is evenly distributed along their length. At this time, the optical cable between the two pole towers is in the shape of a catenary. [2] ,like Figure 5 shown.

[0093] In the figure, A and B are two suspension points, and the catenary equation is shown in formula (1).

[0094]

[0095] Where l is the distance between the two towers; h is the height difference between the two suspension points A and B; σ0 is the horizontal stress at the lowest point of the sag; and γ is the specific load of the optical cable.

[0096] Since the catenary equation contains hyperbolic functions, the calculation is relatively complicated, and the length of the OPGW optical cable differs from the slant span (the distance between the two suspension points) by about a thousandth, the catenary equation is usually simplified to an oblique parabola equation, which can be expressed as:

[0097]

[0098] In the formula, Indicates the angle between the slant pitch and the x-axis.

[0099] The horizontal stress σ0 at any point of the overhead line in the section is equal everywhere, and the stress σ x It refers to the axial stress at this location and can be expressed as:

[0100]

[0101] The stress magnitudes at suspension points A and B are shown in formulas (4) and (5).

[0102]

[0103]

[0104] Assume that the line span is 800m, the height difference is 20m, the horizontal stress σ0 at the lowest point of the sag is 95MPa, and the specific load γ is 60×10 -3 MPa / m, the maximum strain value of the optical fiber at the top of the right tower is 0.1%. According to formula (3), the strain distribution of the OPGW optical cable core under the catenary strain can be obtained, as follows: Figure 6 shown.

[0105] The fusion system prototype was used to test a 500kV OPGW optical cable in the Three Gorges area. This line is located in a mountainous area, passing through a large elevation difference, a large span, and a heavy ice area. The optical cable is affected by the terrain and climate, and the excess length is released quickly. The strain of the fiber core of multiple sections of the optical cable is in the form of a catenary. The test results show that the entire tension section produces a strain of about 0.1%, and the optical fiber strain between adjacent towers is similar to the catenary shape. The maximum strain point is located at the top of the tower, and its strain shape is consistent with the catenary shape.

[0106] Since the OPGW optical cable is a loose tube optical cable, the fiber core can move freely in the optical unit under the action of grease and excess length, and the optical fiber does not produce strain before the excess length is exhausted. When the optical cable is in the form of a catenary, the internal optical fiber does not necessarily produce strain of the same form. At the same time, the complexity of the cause of the stress of the optical cable, the load usually received by the optical cable is non-uniform, and the stresses between the continuous gears affect each other, the influence of bending strain and alternating stress, so that the internal optical fiber is stressed in different states, and the strain does not fully meet the catenary theory formula. Therefore, the present invention is subsequently subjected to stress analysis of the optical cable in the form of strain, and no quantitative calculation is performed.

[0107] 3) Standing wave morphology analysis

[0108] OPGW optical cables are not only subject to vertical forces caused by their own weight and ice weight, but also to horizontal forces caused by wind pressure. The horizontal forces caused by wind pressure on optical cables are affected by many factors, including wind speed, cable windward area, and the angle between the cable and wind direction. Wind, ice and temperature are the three meteorological factors that affect power transmission lines. The effect of wind, in addition to generating horizontal loads in the vertical line direction acting on conductors, ground wires and towers, is also the fundamental reason for the vibration and dancing of OPGW optical cables.

[0109] The OPGW optical cable exhibits a catenary shape under the action of its own gravity, as shown in formula (2). The optical cable will vibrate under the action of wind. Long-term stable wind vibration will cause fatigue and deformation of the optical cable, causing standing wave strain in the optical cable core, as shown in Figure 7 shown.

[0110] OPGW optical cables will vibrate under the influence of external wind. When the wind matches the resonant frequency of the optical cable structure, the vibration of the optical cable will intensify, thus generating wind vibration. Stable wind vibration generates standing waves along the optical cable, and its expression is:

[0111] y=Acos2π(x / λ)cos2π(t / T) (6)

[0112] Where λ is the standing wave wavelength and T is the standing wave period.

[0113] Long-term periodic vibration will cause plastic deformation of the material, resulting in standing wave strain in the internal optical fiber. Under the action of standing wave, the optical cable core will produce strain, and the strain at each point is expressed as the absolute value of the standing wave envelope, as shown in formula (7):

[0114] y=|Acos2π(x / λ)| (7)

[0115] Therefore, the OPGW optical cable core is a superposition of the catenary shape and the standing wave shape under the action of its own gravity and wind vibration, and its expression is:

[0116]

[0117] In the formula, represents the angle between the slant span and the x-axis; l represents the span; σ0 represents the horizontal stress of the overhead line; γ represents the specific load of the optical cable; λ represents the standing wave wavelength; A represents the amplitude of the standing wave signal.

[0118] Assume that the line span is 800m, the height difference is 5m, the horizontal stress σ0 at the lowest point of the sag is 95Mpa, and the specific load γ is 60×10 -3 MPa / m, the standing wave wavelength is 200m, and the maximum strain value of the optical fiber is 0.1%. According to formula (8), the strain distribution of the OPGW optical cable core under the standing wave morphological strain can be obtained by simulation, as follows: Figure 8 shown.

[0119] 4) Local large strain morphology analysis

[0120] In the catenary morphology strain theory analysis, the OPGW cable is assumed to be a non-rigid rope with uniform force, and its force satisfies the catenary equation. The actual OPGW cable is formed by twisting metal monofilaments such as aluminum wire, aluminum-clad steel, and stainless steel tubes. It exhibits rigidity when locally bent, so its force is not exactly the same as the catenary theory. It not only bears axial tension, but also bending stress. [3] .

[0121] Fig. 9 This is the force diagram of the OPGW cable under rigid conditions within the span. Assuming that the specific load acting on the OPGW cable is evenly distributed horizontally along the oblique span, the load concentration on the horizontal projection of the OPGW cable per unit length can be expressed as:

[0122] p0=γA / cosβ (9)

[0123] In the formula, γ represents the specific load of the optical cable; β represents the angle between the line connecting the two suspension points and the horizontal direction; A represents the cross-sectional area of ​​the optical cable.

[0124] The suspension point is subject to horizontal tension and vertical reaction force R A , R B and the restraining moment M A 、M B The horizontal component of tension is T0 and is equal everywhere. Fig. 9 In the selected coordinate system, the bending moment M of the rigid OPGW cable with bending strength EJ at any point x away from the low suspension point A is x and the bending moment M at any point x′ from the high suspension point B x′ They are:

[0125]

[0126]

[0127] In the formula, From the above two equations, it can be seen that the suspension point bends and generates bending moment under the clamping of the hardware. The bending moment is the largest at the suspension point. As it moves away from the suspension point, the bending moment of the optical cable decays exponentially. The bending stress generated by the bending moment is mainly concentrated near the suspension point of the optical cable, which manifests as a large local strain at the top of the tower.

[0128] Assuming that the span between the two sections of the line is 400m, the height difference is 10m, and the horizontal tension T0 at the lowest point of the sag is 35kN, the EJ obtained from the experiment is 145MN·mm 2 , the load concentration p0 on the horizontal projection of the optical cable is 18N / m, and the maximum strain value of the optical fiber is 0.5%. According to formula (11), the strain distribution of the OPGW optical cable under the action of local large strain can be obtained by simulation, as shown in Fig.10 shown.

[0129] Then, S4 is executed. In S4, the core strain morphology type is used as a classification label and input into a deep learning model together with the core strain data sequence for training to obtain a trained deep learning-based classification model.

[0130] According to an embodiment of the present invention, the deep learning model is a convolutional neural network (CNN). CNN is a neural network containing a convolutional layer (CL), which has been widely used in multiple fields such as image processing, natural language processing, recommendation systems and speech recognition. Its network consists of an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer. The convolution layer convolves the input data with the convolution kernel, thereby extracting features from the input data and removing redundant data and noise. Generally, the convolution layer contains multiple convolution kernels, which can extract data features from different dimensions, enrich the useful information of the data, and facilitate model training. Since the convolution operation belongs to linear calculation, the ReLU activation function should be added after the convolution layer to increase the nonlinear ability of the model. The data after the convolution layer will increase significantly. In order to reduce redundant data and model calculation, it is necessary to add a pooling layer (PL) after the convolution layer. Common pooling operations include maximum pooling and average pooling. The input matrix data is divided into several rectangular areas, and the maximum value or average value is calculated for each area to form a new matrix after pooling. The new matrix retains as much feature information of the original matrix as possible and removes a large amount of redundant information. However, too many pooling operations will cause information loss. The pooling layer needs to be optimized according to the characteristics of the input data and the model structure. After the pooling operation, the output data needs to be flattened in one dimension as the input of the fully connected layer (FCL), and finally the classification result is given in the output layer. CNN can be regarded as convolution and pooling of the input data, extracting feature information and then training the model.

[0131] In the data set, the cable segments with the maximum core strain value within the span less than 0.05% are marked as zero strain. For the cable segments with strain exceeding 0.05%, they are marked as catenary, standing wave or local large strain according to the cable strain shape. The typical strain waveform is as follows Fig.11 As shown. The zero strain form is that the maximum strain of the optical cable segment is less than 0.05%, and the core strain form is not considered; the catenary form is that the strain of the optical cable segment satisfies the catenary equation, including catenaries of equal height and catenaries of unequal height; the standing wave form is that the strain in the optical cable segment is a periodic standing wave form, which may also be the superposition of standing waves and catenaries; the local large strain form is that there is a short-distance strain mutation in the optical cable segment, and the maximum strain point is generally located near the suspension point.

[0132] After the fiber core strain data sequences of the plurality of optical cable segments are acquired, they need to be preprocessed. The preprocessing includes: performing a truncation or zero-filling operation on the fiber core strain data sequence of each optical cable segment.

[0133] Specifically, the distribution of each strain type in the strain core of the OPGW optical cable in the statistical data set is 8895 zero strain morphology cable segment data, 879 catenary morphology cable segment data, 667 standing wave morphology cable segment data, and 183 local large strain morphology cable segment data. If all the strain data of the optical cable segments are used for model training, the model has a good classification effect on zero strain due to the high proportion of zero strain data, while the classification errors of the other three types of strain are large. In order to balance the proportion of various strain types and consider the actual proportion of various types of strain in the actual OPGW optical cable line, 2000 pieces of zero strain morphology data are randomly selected from 8895 pieces of zero strain morphology data, and used as the data set of the CNN model together with all the strain data of the other three forms. Since the length of the strain data of each optical cable segment is inconsistent, the strain data of the optical cable segment needs to be truncated or zero-filled when constructing the data set. The maximum length of each strain type data is statistically: 4449 for zero strain morphology, 2480 for catenary morphology, 2537 for standing wave morphology, and 206 for local large strain morphology. Since the maximum data length of the zero strain morphology is significantly larger than that of the other three strain types, and the zero strain data contains less information, the data set length is selected as 2560, the data exceeding the data set length is truncated, and the data less than the data set length is padded with zeros, and finally constitutes the CNN model data set together with the strain type.

[0134] The CNN model is used to classify the optical cable core strain data. Its structure is as follows: Fig.12As shown in the figure. After the strain data enters the CNN model, it passes through three convolution modules, each of which consists of a convolution layer, a ReLU activation function, and a maximum pooling layer. For a one-dimensional strain sequence with a length of 2560, after 32 cores (1×3), 64 cores (1×3), 128 cores (1×3) and corresponding three maximum pooling layers (2×2), the output data size is (128×320). After being flattened into a one-dimensional sequence, it enters the fully connected layer, which contains two hidden layers with 256 and 64 nodes respectively. The final output layer contains 4 nodes, corresponding to 4 strain types.

[0135] Then, S5 is executed. In S5, the optical cable core strain data sequence corresponding to the OPGW optical cable to be tested, which is collected by the BOTDA system or the BOTDR system, is input into the trained deep learning-based classification model to obtain the core strain morphology type corresponding to each optical cable segment in the OPGW optical cable to be tested.

[0136] S6 is further executed. In S6, after the core strain morphology type corresponding to each optical cable segment in the OPGW optical cable to be tested is obtained, the following analysis and processing are performed: if the core strain morphology type corresponding to a certain optical cable segment is a catenary morphology, then according to the strain value in the core strain data sequence and in combination with the temperature, it is determined whether ice is applied to the optical cable segment; if the core strain morphology type corresponding to a certain optical cable segment is a standing wave morphology, then according to the strain value in the core strain data sequence and in combination with the field observation results, it is further determined whether loose strands and twisted wire breakage occur in the optical cable segment; if the core strain morphology type corresponding to a certain optical cable segment is a local large strain morphology, then according to the field observation results, it is further determined whether the twisted wire at the clamping position of the tower top wire clamp corresponding to the optical cable segment is deformed.

[0137] According to an embodiment of the present invention, for each strain form of the OPGW optical cable to be tested, its application in OPGW optical cable fault diagnosis is analyzed.

[0138] 1) For the catenary shape: the 60#~69# pole tower of a 500kV OPGW optical cable in central China was selected for analysis. When there is no icing, the strain of the optical cable in the tension section is uniform, and the overall strain value is around 0.05%, which is basically not subject to strain. When icing occurs in winter, the optical cable is stretched due to the increase in gravity after icing. After the excess length of the fiber core inside the optical unit is exhausted, the optical fiber and the optical cable are deformed together, and the optical fiber strain is in the shape of a catenary. The entire tension section produces 10 stress maximum points, corresponding to the 10 pole towers from 60# to 69#, with a maximum strain value of 0.1%, which is significantly increased compared to the non-icing period. The icing state of the OPGW optical cable can be identified based on the strain value of the optical cable core and the strain shape of the optical cable in the tension section, combined with the comprehensive analysis of the meteorological information of the day. The maximum strain value of the top of the 63# tower was selected to analyze the strain change trend within a day, and the fusion system prototype was used for online monitoring, with a measurement interval of 30 minutes. The strain of the optical cable core was less than 0.05% during the period of 0:00 to 8:00, and no ice was formed on the OPGW optical cable during this period. The strain of the optical cable core gradually increased during the period of 8:00 to 9:30, and reached the maximum strain of 0.1% at 9:30. Due to the increase in temperature or sunlight during the period of 9:30 to 11:30, the ice gradually disappeared, and returned to the pre-icing state after 11:30. The fusion system prototype was used to monitor the strain of the OPGW optical cable core online. According to the strain value and strain morphology of the optical cable core, it can be judged whether the optical cable is iced, and the entire process of the generation, continuation and disappearance of the optical cable ice can be analyzed according to the strain change trend, which is helpful to analyze the impact of ice on the OPGW optical cable core.

[0139] 2) For standing wave morphology: When testing the strain of a 500kV OPGW optical cable in Liaoning, it was found that there was an abnormal strain area in a tension section of the line, and there was no obvious strain in the optical cable cores of several adjacent tension sections before and after the tension section. The method of connecting the tower positioning was used to find that the strain abnormal area was located between the 164# and 166# towers, and the maximum strain value was close to 0.1%. In the measurement results of the strain abnormal area, the strain of the optical cable core showed periodic changes. The strain of the optical cable section between the 164# and 166# towers was based on the unequal height catenary shape and superimposed on the periodic strain caused by wind vibration. The strain results were consistent with the strain characteristics of the standing wave morphology. Due to the fixing effect of the tower top hardware on the optical cable, the optical cable is always a wave node at the top of the tower during the vibration process, and the strains at the tops of the 165# and 166# towers are the two minimum values ​​of the optical cable section.

[0140] A drone was used to conduct an on-site survey of the strain anomaly area of ​​the 164#~166# towers. According to the tower number, tower type, span, height difference, etc. in the line tower list, combined with the geographical information of the area, a cable topographic map was drawn. The optical cable in the strain anomaly area is located at the mountain pass. This geographical feature makes the wind force on the optical cable relatively concentrated and large. At the same time, the 165#~166# towers belong to the area with large height difference and large span, and there are strong winds all year round. The wind acts vertically on the optical cable along the mountain pass, which easily causes wind vibration of the optical cable. There is a wind farm nearby, and the wind direction is perpendicular to the direction of the optical cable. The wind direction perpendicular to the direction of the optical cable is also conducive to the vibration of the optical cable. When designing the line, a shock-proof hammer has been added near the top of the 166# tower to reduce the impact of wind vibration on the OPGW optical cable. However, during the construction and actual operation, the horizontal tension of the optical cable, the impact of the nearby wind farm, and the changes in wind direction and wind speed will change the vibration frequency of the optical cable caused by wind vibration, causing the designed shock-proof measures to fail.

[0141] The OPGW optical cables at the top of several towers near the 166# tower were photographed using drones to observe the impact of abnormal strain caused by wind vibration on the optical cables. Photos of the optical cables at the top of the towers in the normal strain area nearby showed that the OPGW optical cables at the top of the tower were twisted normally, without loose strands or broken wires. The enlarged image of the top of the 166# tower showed that the OPGW optical cables on the left side of the wire clamp had loose strands and broken strands. This is because continuous and periodic vibrations can easily fatigue the optical cable material, and the optical cables at the top of the tower are located at the wave node. The optical cables at this point do not vibrate, while the optical cables on both sides continue to vibrate, which is more likely to cause fiber breakage, strand breakage, and wire breakage accidents in the OPGW optical cables at the top of the tower. The wind vibration of the OPGW optical cable can also be measured using distributed vibration monitoring equipment, and the frequency of the vibration signal can be analyzed to determine whether there is a breeze vibration or dancing event. Using the fusion system prototype to monitor the standing wave morphological strain of the OPGW optical cable can analyze the fatigue stress caused by periodic vibration of the optical cable, more intuitively reflect the impact of wind vibration on the optical cable, and have more practical application value in line status assessment and optical cable operation and maintenance.

[0142] 3) For local large strain forms: The BOTDA mode of the fusion system prototype was used to test a cross-provincial OPGW optical cable line in Northeast China. It was found that there was a local large strain area 64.2km away from the test point. The strain abnormality area was located using the splicing tower positioning method, and it was determined that the strain abnormality area was located at the top of the 384# straight tower. The strain section where the strain abnormality area is located includes 12 towers from 377# to 388#. Except for the local strain abnormality at the top of the 384# tower, the strain in other areas of the entire tension section is less than 0.05%, and there is no obvious strain. The length of the strain area is 50m, and the maximum strain value is 0.522%, which belongs to the local large strain form. Its strain form is significantly different from the catenary strain and standing wave strain. According to the results of the on-site survey, the terrain of the tension section is flat, and the height of the tower in the tension section is between 30 and 36m, and there is no stress caused by the large height difference on the OPGW optical cable core. Since the strain is within 50m of the top of the 384# straight tower, preliminary analysis shows that the strain on the optical cable core is caused by abnormal hardware at the top of the tower. Using drones to photograph the hardware at the top of several towers near 384#, and comparing the normal 383# tower with the 384# tower with local strain, it can be seen that the strands at the top of the 384# tower are unevenly distributed. This is because the clamping force of the wire clamp is too large, causing the strands to deform at the clamping point, generating bending moment stress in the optical cable, and causing large local strain at the top of the tower.

[0143] OPGW optical cable core strain tests were conducted in typical areas across the country, including plains, mountains, plateaus, high cold, strong winds, ice and other natural and meteorological conditions. The strain of the optical cable was analyzed and counted for each span in combination with the environment in which the optical cable is located, and the following conclusions were obtained:

[0144] Using the long-distance fusion system prototype and strain demodulation technology, the Brillouin frequency shift test results are converted into cable core strains for each span. Combined with the theoretical threshold of the optical cable and the actual fault situation, the optical cable segment with a strain value less than 0.05% is treated as zero strain. The strain data of all optical cable segments are statistically analyzed, and the strain is divided into four categories according to morphology: zero strain morphology, catenary morphology, standing wave morphology and local large strain morphology. The OPGW optical cable core has 0.6% to 0.7% excess length, and due to the effect of the grease, the core can slide freely in the optical cable, and the optical cable is not stressed before the excess length is exhausted. After the excess length of the optical cable core is exhausted, under the action of gravity, its force is manifested in the form of a catenary, which is verified by theoretical simulation and measured data. Under the action of wind vibration, the OPGW optical cable generates standing waves, and the internal core generates standing wave morphological strain. Theoretical simulation and measured data show that the standing wave morphological strain is the superposition of the catenary and the standing wave. Under the action of the tower top clamp, the optical cable core will generate bending stress due to the bending moment. Since the bending stress decays exponentially, the bending stress drops sharply away from the suspension point, which is manifested as a local large strain form at the top of the tower. For newly built OPGW optical cable lines, the line construction quality is evaluated by measuring whether the strain of the optical cable core exceeds the zero strain threshold. The weight of the optical cable increases after being covered with ice. When the excess length of the internal core is exhausted, it will show a catenary strain. The ice condition of the optical cable can be judged by the strain value and strain form. Long-term wind vibration will cause the optical cable core to produce standing wave strain. The impact of wind vibration on the optical cable and the core can be judged based on the strain value and form. The local large strain form is related to the tower top clamp and the optical cable strands. The safety status of the tower top clamp and the strands can be judged based on the local point strain value and form.

[0145] Another embodiment of the present invention provides an OPGW optical cable core strain classification system based on supervised machine learning, the system comprising:

[0146] A signal acquisition module configured to acquire the Brillouin frequency shift of the OPGW optical cable using a BOTDA system or a BOTDR system;

[0147] A signal processing module, configured to obtain a core strain data sequence of each optical cable segment according to the Brillouin frequency shift demodulation; analyze and process the core strain data sequence to obtain a core strain morphology type corresponding to each optical cable segment; wherein the core strain morphology type includes a zero strain morphology, a catenary morphology, a standing wave morphology, and a local large strain morphology;

[0148] A model training module is configured to input the core strain morphology type as a classification label and the core strain data sequence into a deep learning model for training, thereby obtaining a trained classification model based on deep learning;

[0149] The strain morphology classification module is configured to input the optical cable core strain data sequence corresponding to the OPGW optical cable to be tested, which is collected by the BOTDA system or the BOTDR system, into the trained deep learning-based classification model to obtain the core strain morphology type corresponding to each optical cable segment in the OPGW optical cable to be tested.

[0150] In this embodiment, preferably, the signal processing module analyzes and processes the core strain data sequence to obtain the core strain morphology type corresponding to each optical cable segment, including: if each strain value in the core strain data sequence is less than a preset threshold, it indicates that the optical fiber is not stressed, and the core strain morphology type corresponding to the optical cable segment is a zero strain morphology;

[0151] If the strain distribution corresponding to the core strain data sequence conforms to the catenary equation, and the maximum strain value exceeds the preset threshold, the core strain morphology type corresponding to the optical cable segment is the catenary morphology;

[0152] If the strain distribution corresponding to the core strain data sequence is in the form of a periodic standing wave or conforms to the superposition equation of the catenary form and the standing wave form, and the maximum strain value exceeds the preset threshold, then the core strain form type corresponding to the optical cable section is the standing wave form;

[0153] If a local strain value that is short-lived and far higher than a preset threshold corresponding to a zero strain morphology appears in the strain distribution corresponding to the core strain data sequence, the core strain morphology type corresponding to the optical cable segment is a local large strain.

[0154] The function of the OPGW optical cable core strain classification system based on supervised machine learning described in this embodiment can be described by the aforementioned OPGW optical cable core strain classification method based on supervised machine learning. For the parts not described in detail in this embodiment, please refer to the above method embodiment.

[0155] It should be noted that, although several units, modules or submodules are mentioned in the above detailed description, this division is merely exemplary and not mandatory. In fact, according to an embodiment of the present invention, the features and functions of two or more modules described above can be embodied in one module. Conversely, the features and functions of one module described above can be further divided into being embodied by multiple modules.

[0156] In addition, although the operations of the method of the present invention are described in a specific order in the drawings, this does not require or imply that the operations must be performed in this specific order, or that all the operations shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

[0157] Although the spirit and principle of the present invention have been described with reference to several specific embodiments, it should be understood that the present invention is not limited to the disclosed specific embodiments, and the division of various aspects does not mean that the features in these aspects cannot be combined to benefit, and such division is only for the convenience of expression. The present invention is intended to cover various modifications and equivalent arrangements included in the spirit and scope of the attached claims.

[0158] The documents cited in the present invention are as follows:

[0159] [1] Chen Xihao, Huang Junhua, Zhang Jianming, et al. Analysis and suggestions on OPGW overload caused by ice disaster [J]. Power System Communications, 2008, 29(9): 8-13.

[0160] [2] Meng Suimin, Kong Wei. Design of Overhead Transmission Lines[M]. China Electric Power Press, 2007.

[0161] [3] Bao Yuben, Sun Junqiang, Huang Qiang. Research progress of Brillouin optical time domain reflectometer distributed fiber optic sensing[J]. Laser & Optoelectronics Progress, 2020, 57(21): 210002-1.

Claims

1. An OPGW optical cable core strain classification method based on supervised machine learning is characterized in that: include: Use BOTDA system or BOTDR system to collect Brillouin frequency shift of OPGW optical cable; Acquire a core strain data sequence of each optical cable segment according to the Brillouin frequency shift demodulation; Analyze and process the core strain data sequence to obtain the core strain morphology type corresponding to each optical cable segment; wherein the core strain morphology type includes zero strain morphology, catenary morphology, standing wave morphology, and local large strain morphology; The fiber core strain morphology type is used as a classification label and input into a deep learning model together with the fiber core strain data sequence for training to obtain a trained classification model based on deep learning; The optical cable core strain data sequence corresponding to the OPGW optical cable to be tested, collected by the BOTDA system or the BOTDR system, is input into the trained classification model based on deep learning to obtain the core strain morphology type corresponding to each optical cable segment in the OPGW optical cable to be tested.

2. The OPGW optical cable core strain classification method based on supervised machine learning according to claim 1, characterized in that: The method of obtaining the core strain data sequence of each optical cable segment according to the Brillouin frequency shift calculation includes: obtaining multiple optical cable segments of different lengths, including: assuming that the core length of the optical cable in the tension section is L, the heights of the tension towers on both sides are h1 and h2, and the down conductor is reserved for welding, then the core length of the optical cable in the tension section after removing the down conductor is L fiber = L-h1-h2-2h; Determine the number of connected towers n and the spacing l between two adjacent towers in the tension section according to the tower list i , i=1,2,…,n-1, calculate the cumulative span in the tension section Calculate the ratio of the optical cable core length to the cumulative span length k = L fiber / L tower , the core length of the optical cable is L i =k×l i , thereby obtaining multiple optical cable segments of different lengths; based on the different lengths of the divided optical cable segments, the Brillouin frequency shift is demodulated to obtain the core strain data sequences of the multiple optical cable segments.

3. The OPGW optical cable core strain classification method based on supervised machine learning according to claim 1, characterized in that: The analyzing and processing the fiber core strain data sequence to obtain the fiber core strain morphology type corresponding to each optical cable segment includes: If each strain value in the fiber core strain data sequence is less than the preset threshold, it indicates that the optical fiber is not stressed, and the fiber core strain morphology type corresponding to the optical cable segment is zero strain morphology; If the strain distribution corresponding to the core strain data sequence conforms to the catenary equation, and the maximum strain value exceeds the preset threshold, the core strain morphology type corresponding to the optical cable segment is the catenary morphology; If the strain distribution corresponding to the core strain data sequence is in the form of a periodic standing wave or conforms to the superposition equation of the catenary form and the standing wave form, and the maximum strain value exceeds the preset threshold, then the core strain form type corresponding to the optical cable section is the standing wave form; If a local strain value that is short-lived and far higher than a preset threshold corresponding to a zero strain morphology appears in the strain distribution corresponding to the core strain data sequence, the core strain morphology type corresponding to the optical cable segment is a local large strain.

4. The OPGW optical cable core strain classification method based on supervised machine learning according to claim 3 is characterized in that: The catenary equation is: Where y represents strain, x represents the position of the optical cable, It represents the angle between the slant span and the x-axis, l represents the span between the two towers; σ0 represents the horizontal stress at the lowest point of the sag; γ represents the specific load of the optical cable.

5. The OPGW optical cable core strain classification method based on supervised machine learning according to claim 3 is characterized in that: The superposition equation of the catenary shape and the standing wave shape is: Where y represents strain, x represents the position of the optical cable, represents the angle between the slant span and the x-axis, l represents the span between the two towers; σ0 represents the horizontal stress at the lowest point of the sag; γ represents the specific load of the optical cable; λ represents the standing wave wavelength; and A represents the amplitude of the standing wave signal.

6. The OPGW optical cable core strain classification method based on supervised machine learning according to claim 2, characterized in that: After the fiber core strain data sequence of each optical cable segment is acquired, it is preprocessed. The preprocessing includes: performing a truncation or zero-filling operation on the fiber core strain data sequence of each optical cable segment.

7. The OPGW optical cable core strain classification method based on supervised machine learning according to claim 1, characterized in that: The deep learning model is a convolutional neural network; the core strain morphology type is used as a classification label and input into the deep learning model together with the core strain data sequence for training, including: the core strain data sequence passes through three convolution modules respectively, each convolution module includes a convolution layer, a ReLU activation function and a maximum pooling layer, that is, passes through 32 cores, 64 cores, 128 cores and corresponding 3 maximum pooling layers; then enters the fully connected layer, the fully connected layer includes two hidden layers, and the data output by the maximum pooling layer is flattened into a one-dimensional sequence; the final output layer includes 4 nodes, corresponding to 4 types of core strain morphology.

8. The OPGW optical cable core strain classification method based on supervised machine learning according to claim 1, characterized in that: After obtaining the fiber core strain morphology type corresponding to each cable segment in the OPGW optical cable to be tested, the following analysis and processing are performed: If the fiber core strain morphology type corresponding to a certain optical cable segment is a catenary shape, then the strain value in the fiber core strain data sequence is used to determine whether the optical cable segment is iced in combination with the temperature; If the core strain morphology type corresponding to a certain optical cable segment is a standing wave morphology, then based on the strain value in the core strain data sequence and combined with the field observation results, it is further determined whether the optical cable segment has loose strands and twisted wire breaks; If the strain morphology type of the fiber core corresponding to a certain optical cable segment is a local large strain morphology, then based on the on-site observation results, it is further determined whether the strands at the clamping position of the tower top clamp corresponding to the optical cable segment are deformed.

9. The OPGW optical cable core strain classification system based on supervised machine learning is characterized by: include: A signal acquisition module configured to acquire the Brillouin frequency shift of the OPGW optical cable using a BOTDA system or a BOTDR system; A signal processing module, configured to obtain a core strain data sequence of each optical cable segment according to the Brillouin frequency shift demodulation; analyze and process the core strain data sequence to obtain a core strain morphology type corresponding to each optical cable segment; wherein the core strain morphology type includes a zero strain morphology, a catenary morphology, a standing wave morphology, and a local large strain morphology; A model training module is configured to input the core strain morphology type as a classification label and the core strain data sequence into a deep learning model for training, thereby obtaining a trained classification model based on deep learning; The strain morphology classification module is configured to input the optical cable core strain data sequence corresponding to the OPGW optical cable to be tested, which is collected by the BOTDA system or the BOTDR system, into the trained deep learning-based classification model to obtain the core strain morphology type corresponding to each optical cable segment in the OPGW optical cable to be tested.

10. The OPGW optical cable core strain classification system based on supervised machine learning according to claim 9, characterized in that: The signal processing module analyzes and processes the core strain data sequence to obtain the core strain morphology type corresponding to each optical cable segment, including: if each strain value in the core strain data sequence is less than a preset threshold, it indicates that the optical fiber is not stressed, and the core strain morphology type corresponding to the optical cable segment is a zero strain morphology; If the strain distribution corresponding to the core strain data sequence conforms to the catenary equation, and the maximum strain value exceeds the preset threshold, the core strain morphology type corresponding to the optical cable segment is the catenary morphology; If the strain distribution corresponding to the core strain data sequence is in the form of a periodic standing wave or conforms to the superposition equation of the catenary form and the standing wave form, and the maximum strain value exceeds the preset threshold, then the core strain form type corresponding to the optical cable section is the standing wave form; If a local strain value that is short-lived and far higher than a preset threshold corresponding to a zero strain morphology appears in the strain distribution corresponding to the core strain data sequence, the core strain morphology type corresponding to the optical cable segment is a local large strain.

Images