Communication optical cable state real-time online monitoring method and system

Abstract

The invention belongs to the technical field of communication, and provides a communication optical cable state real-time online monitoring method and system. Calculating the vibration displacement of the optical cable based on the original back scattering light signal, and calculating the real strain and temperature of the optical cable based on a Brillouin scattering signal and a Raman scattering signal obtained by separating the original back scattering light signal; carrying out filtering, denoising, normalization and time-space alignment processing on the vibration displacement of the optical cable, the real strain of the optical cable and the temperature, and binding the vibration displacement and the real strain with the obtained geographic coordinates; obtaining time domain, frequency domain and time-frequency domain features based on the preprocessed vibration displacement, real strain and temperature data to form a multi-dimensional feature vector, inputting the feature vector into a pre-trained event recognition model, and outputting an abnormal event classification result; and judging and triggering abnormal event early warning based on an output result of the event recognition model and a preset dynamic threshold.

Description

Technical Field

[0001] This invention belongs to the field of communication technology, and in particular relates to a method and system for real-time online monitoring of the status of communication optical cables. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] The long-term stable operation of optical fiber cables is directly related to the continuity of communication networks and the reliability of power grid services, playing a vital supporting role in public services and economic development. However, optical fiber networks laid over long distances and large areas are susceptible to multi-dimensional threats, such as those along urban trunk roads, overhead optical cables in mountainous areas, and cross-regional backbone optical cables.

[0004] Regarding external threats, continuous vibration or direct mechanical damage from third-party construction and excavation, deformation of optical cables caused by heavy vehicles running over them, and damage to optical fiber cores caused by animal gnawing can all lead to optical cable breakage or signal transmission interruption. Third-party construction and excavation includes road expansion, pipeline laying, etc.

[0005] Regarding the threats from complex environments, extreme weather conditions can lead to the deterioration of optical cable material performance, water immersion corrosion caused by continuous rainfall or water accumulation, and excessive tensile strain of optical cables caused by soil subsidence or geological activity. These factors can gradually reduce the transmission quality of optical cables and induce latent faults. Extreme weather conditions include high temperatures and severe freezing.

[0006] In terms of long-term operational losses, material aging after long-term service of optical cables, loosening of joints due to failure of joint box sealing, and local stress concentration caused by uneven stress on the line can easily form signal attenuation points or potential faults. Material aging includes sheath cracking, etc.

[0007] Regarding the risk of location deviation, natural factors such as geological landslides and ground subsidence, or unexpected movement of the optical cable route due to surrounding construction, may damage the original laying status of the optical cable and increase the risk of breakage or excessive loss.

[0008] Faced with the aforementioned complex threats, existing mainstream optical cable operation and maintenance monitoring methods include manual inspection, OTDR offline testing, and simple optical power alarm methods. Manual inspection cannot promptly detect sudden faults and hidden risks. OTDR offline testing typically requires interrupting communication services for a specific time, failing to provide continuous, real-time optical cable status data. Its sensitivity to minute strain changes and slow degradation processes such as early material aging is extremely low, making early fault warnings and trend assessments difficult. Simple optical power alarm methods determine continuity or loss levels solely by monitoring whether the link's optical power value falls below a threshold. This method cannot accurately locate fault points, identify specific fault types, or provide early warnings of potential risks such as stress concentration. Therefore, facing these complex threats, existing mainstream optical cable operation and maintenance monitoring methods have significant limitations in adapting to the comprehensive monitoring needs of long-distance, large-scale, and complex scenarios, making it difficult to meet the requirements of high-reliability services such as power grids for real-time perception and risk warning of optical cable status. Summary of the Invention

[0009] To address at least one of the technical problems mentioned above, this invention provides a method and system for real-time online monitoring of the status of communication optical cables. This system meets the comprehensive monitoring needs of complex scenarios and satisfies the requirements of high-reliability services such as power grids for real-time perception and risk warning of optical cable status.

[0010] To achieve the above objectives, the present invention adopts the following technical solution: The first aspect of the present invention provides a method for real-time online monitoring of the status of communication optical cables, comprising the following steps: The probe light pulse is transmitted into the optical fiber of the communication cable, which serves as the sensing medium, and the backscattered light signal is received and demodulated. The vibration displacement of the optical cable is calculated based on the original backscattered light signal, and the actual strain and temperature of the optical cable are calculated based on the Brillouin scattering signal and Raman scattering signal obtained by separating the original backscattered light signal. The vibration displacement, actual strain, and temperature of the optical cable are filtered and denoised, normalized, and spatiotemporally aligned respectively. The vibration displacement and actual strain are then bound to the acquired geographic coordinates. Based on the preprocessed vibration displacement, real strain and temperature data, time domain, frequency domain and time frequency domain features are extracted to form a multi-dimensional feature vector. The feature vector is then input into a pre-trained event recognition model to output the abnormal event classification result. Based on the output of the event recognition model and the preset dynamic threshold, abnormal event warnings are determined and triggered.

[0011] Furthermore, the calculation of the true strain and temperature of the optical cable based on the Brillouin scattering signal and Raman scattering signal obtained by separating the original backscattered light signal includes: Wavelet threshold filtering was performed on the acquired Raman scattering signals to extract the Stokes / anti-Stokes intensity peaks at the same spatial location. , The absolute temperature of the optical cable is solved by inverse calculation, and the temperature deviation is calculated. The true temperature of the optical cable is then calculated. By combining the temperature deviation, the pure strain frequency shift component is calculated based on the detected Brillouin scattering signal. The true strain parameters are then obtained by combining the pure strain frequency shift component with the compensation formula.

[0012] Furthermore, when binding vibration displacement, actual strain, and acquired geographic coordinates, the process includes timestamping the actual strain data with vibration displacement, temperature, and geographic coordinate data; establishing a "fiber optic length-geographic coordinate" mapping relationship based on the acquired location information of key monitoring nodes; generating a spatial distribution topology map of the optical cable; and binding the actual strain data of each monitoring point with the corresponding longitude, latitude, and optical cable segment ID based on the optical cable spatial distribution topology map.

[0013] Furthermore, the multi-dimensional feature vector is formed by extracting time-domain, frequency-domain, and time-frequency-domain features from the preprocessed vibration displacement, true strain, and temperature data, including: Time-domain features include vibration time-domain features and strain time-domain features. Vibration time-domain features include peak amplitude of vibration signal and vibration energy, while strain time-domain features include strain gradient and strain rate of change. The frequency domain features are generated based on vibration displacement data through fast Fourier transform and are used to distinguish the frequency characteristics of different types of vibration events. The time-frequency domain features are extracted by combining the temporal changes of vibration displacement with abrupt temperature changes, and by capturing the parameter correlation patterns in the dynamic evolution of events. By integrating time-domain features, frequency-domain features, and time-frequency-domain features, a multi-dimensional feature vector is constructed.

[0014] Furthermore, in the CNN-LSTM hybrid architecture for event recognition model training, the constructed multi-dimensional feature vectors are classified, including vibration feature groups, strain-temperature feature groups, and spatiotemporal correlation feature groups. These are then input into different layers of the CNN-LSTM hybrid architecture. The vibration feature group is input into the first convolutional layer of the CNN to capture high-frequency vibration features, while the strain-temperature feature group is input into the second convolutional layer of the CNN to capture the coordinated change features of strain and temperature. After being concatenated with the spatiotemporal correlation features, these features are input into the LSTM. The probability distribution of the event is obtained based on the hidden state output by the LSTM. The parameters of the CNN and LSTM are then adjusted according to the probability distribution of the event.

[0015] Furthermore, the method also includes calculating the health score of the optical cable based on historical real strain data; and using a time-series prediction model based on historical data to predict future health trends and changes in real strain, triggering a preventive maintenance warning when the prediction results meet preset conditions.

[0016] Furthermore, the health rating formula for optical cables is as follows: , in, For health, , These are the weighting coefficients. D represents the number of abnormal events, and D represents the number of monitoring days. For realistic response The cumulative duration for which ε exceeds the preset threshold. Total monitoring duration This represents the business impact coefficient.

[0017] A second aspect of the present invention provides a real-time online monitoring system for the status of communication optical cables, comprising: The signal acquisition and demodulation module is configured to transmit probe light pulses to the optical fiber of the communication optical cable, which serves as the sensing medium, and to receive and demodulate its backscattered light signals. The data demodulation module is configured to calculate the vibration displacement of the optical cable based on the original backscattered light signal, and to calculate the actual strain and temperature of the optical cable based on the Brillouin scattering signal and Raman scattering signal obtained by separating the original backscattered light signal. The data association module is configured to perform vibration displacement, actual strain and temperature filtering and noise reduction, normalization and spatiotemporal alignment processing on the optical cable, and bind the vibration displacement and actual strain to the acquired geographic coordinates. The abnormal event classification module is configured to extract time-domain, frequency-domain, and time-frequency-domain features based on preprocessed vibration displacement, real strain, and temperature data to form a multi-dimensional feature vector. The feature vector is then input into a pre-trained event recognition model to output the abnormal event classification result. The early warning module is configured to determine and trigger early warnings for abnormal events based on the output of the event recognition model and preset dynamic thresholds.

[0018] A third aspect of the present invention provides a computer-readable storage medium.

[0019] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the real-time online monitoring method for the status of a communication optical cable as described above.

[0020] A fourth aspect of the present invention provides a computer device.

[0021] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps of the real-time online monitoring method for the status of a communication optical cable as described above.

[0022] Compared with the prior art, the beneficial effects of the present invention are: This invention utilizes a multi-physical quantity synergistic analysis to simultaneously monitor four core physical quantities: vibration, strain, temperature, and geographical location. Through multi-parameter correlation analysis, it provides more comprehensive information on the optical cable's status. For example, combining vibration and strain data can accurately identify crushing faults, while combining temperature and strain data can effectively distinguish between environmental influences and external force damage, quickly pinpointing the root cause of the fault.

[0023] This invention relies on distributed optical fiber sensing technology to achieve 24 / 7 uninterrupted online monitoring, covering the entire optical cable route. It can instantly capture sudden events such as construction excavation and vehicle crushing (response time ≤ 1 second), and also continuously track slow deterioration processes such as soil subsidence and material aging. It completely solves the problems of long manual inspection cycles and the need to interrupt business and lack of real-time performance for OTDR testing.

[0024] This invention utilizes a CNN-LSTM hybrid architecture-based event recognition model that automatically extracts multi-dimensional features to achieve accurate automatic classification of typical events. Furthermore, by integrating multi-source information such as maintenance plans and meteorological data for collaborative analysis, it effectively suppresses false alarms and significantly improves operational efficiency. Additionally, it constructs a fiber optic cable health assessment model based on long-term monitoring data and combines it with an LSTM trend prediction model to analyze the long-term trends of key parameters such as strain. This allows for the early identification of potential risks such as stress concentration and accelerated aging, thereby shifting the operational model from passive "post-fault repair" to proactive "pre-fault prevention," effectively reducing operational costs and the risk of business interruption.

[0025] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0026] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0027] Figure 1 This is a flowchart of a method for real-time online monitoring of the status of communication optical cables provided in an embodiment of the present invention; Figure 2 This is a flowchart of a real-time online monitoring process for the status of a communication optical cable provided in an embodiment of the present invention. Detailed Implementation

[0028] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0029] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0030] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0031] Example 1 like Figure 1 and Figure 2 As shown in the figure, this embodiment provides a method for real-time online monitoring of the status of communication optical cables, including the following steps: Step 1: Transmit a probe light pulse into the optical fiber of the communication cable, which serves as the sensing medium, receive and demodulate its backscattered light signal, and obtain the raw scattering spectrum data distributed along the length of the optical fiber. In this embodiment, a dedicated sensing fiber deployed in the communication optical cable is used, or a spare optical fiber in the optical cable is directly reused as the sensing medium, eliminating the need for additional sensors and reducing deployment costs.

[0032] Distributed fiber optic sensing technology is used as the core monitoring method. The fiber optic monitoring equipment emits probe light pulses with specific parameters into the sensing fiber. The pulse width is set to 10ns-100ns according to the monitoring accuracy requirements, and the sampling frequency is adjustable from 2kHz to 10kHz. The pulse width of the probe light pulse is adapted to the type of sensing fiber, as shown in Table 1. Table 1 "Fiber Type - Pulse Width - Spatial Resolution" Compatibility Table

[0033] When an optical pulse propagates in an optical fiber, it generates backscattered optical signals carrying information about the surrounding environment due to Rayleigh scattering, Brillouin scattering, or Raman scattering effects. Rayleigh scattering signals are used for subsequent vibration parameter demodulation, Brillouin scattering frequency shift changes are used for strain and temperature parameter demodulation, and Raman scattering signals can help verify the accuracy of temperature data.

[0034] The monitoring equipment receives and demodulates backscattered light signals through a bidirectional coupler and a photodetector. The photodetector includes an amplifier and an analog-to-digital converter, ultimately acquiring continuous raw scattering spectrum data along the fiber length with a spatial resolution of 1-5 meters, meeting the needs of fine monitoring of long-distance optical cables.

[0035] Step 2: Calculate the vibration displacement of the optical cable based on the original backscattered light signal, and calculate the actual strain and temperature of the optical cable based on the Brillouin scattering signal and Raman scattering signal obtained by separating the original backscattered light signal. Specifically, the steps include the following: Step 201: Based on the Φ-OTDR principle, calculate the vibration displacement ΔL of the optical cable based on the original backscattered light signal. Based on the vibration displacement ΔL, calculate the vibration parameters of each monitoring point along the optical cable, including frequency, amplitude and duration. In this embodiment, the optical fiber monitoring device emits probe light with a pulse width of 10ns-100ns (wavelength λ is determined by the device configuration, such as 1550nm). When the optical signal propagates in the sensing fiber, it generates backscattered Rayleigh light. The scattered light is transmitted to a photodetector via a bidirectional coupler, and the phase change of the backscattered Rayleigh light is calculated using a phase demodulation algorithm (such as coherent demodulation). This value directly reflects the phase shift of the optical signal caused by the vibration of the optical cable. It represents the phase change of the backscattered Rayleigh light. The calculation formula is: , in, Let n be the phase change value of the backscattered Rayleigh light, and n be the refractive index of the fiber. To detect the wavelength of light; Will n Substitute into the formula By mathematically transforming and inversely deducing the vibration displacement ΔL, the calculation formula is as follows: For example, when =πrad, λ=1550nm, n=1.468, m (i.e., 261nm) is the length change of a certain monitoring point of the optical cable due to vibration.

[0036] In this embodiment, based on the real-time change data of ΔL, the frequency of the vibration signal is further calculated (through periodic analysis of the change of ΔL over time), the amplitude (the difference between the maximum and minimum values ​​of ΔL), and the duration (the duration of ΔL exceeding the normal threshold). Finally, the vibration parameters of each point along the optical cable are obtained, with a frequency range of 0.1Hz-2kHz and an amplitude range corresponding to the vibration intensity after ΔL conversion (such as the displacement range equivalent to a 0-10mV voltage signal).

[0037] Step 202: Based on the BOTDA / BOTDR principle, strain and temperature information are demodulated by analyzing the Brillouin divergence radio frequency shift change; In this embodiment, the frequency shift range is set to 10GHz-12GHz. A dual-parameter demodulation technique is used to separate the cross-sensitivity effects of strain and temperature. The dual-parameter demodulation technique includes the introduction of Raman scattering temperature compensation, etc. The relationship between Brillouin frequency shift and strain and temperature satisfies the following formula: , in, For frequency shift changes, These are inherent characteristic parameters of optical fibers, provided by the fiber manufacturer or determined through standard strain calibration experiments. The change in optical cable strain, i.e., the degree of deformation of the optical cable, is calculated after separating the cross-sensitivity using dual-parameter demodulation technology. Temperature coefficient, an inherent characteristic parameter of optical fiber, is provided by the optical fiber manufacturer or determined through standard temperature calibration experiments. Temperature deviation, i.e., the difference between the actual temperature of the optical cable and the reference temperature, ensures the accuracy of strain measurement ± Temperature measurement accuracy ±0.5℃.

[0038] Specifically, the steps include the following: Step 2021: Perform wavelet threshold filtering on the acquired Raman scattering signal to extract the Stokes / anti-Stokes light intensity peaks at the same spatial location. , The absolute temperature of the optical cable is determined by inverse analysis, and the temperature deviation is calculated. The temperature sensitivity principle of Raman scattering: Utilizing the characteristic that the intensity ratio of the "Stokes component" and "anti-Stokes component" of Raman scattering in an optical fiber changes with temperature, their relationship satisfies the formula: , in: To counteract Stokes light intensity, is Stokes intensity; A is the inherent proportionality coefficient of the optical fiber, which is determined by the fiber material and obtained through factory calibration. (1550nm) (1450nm) represents the Stokes / anti-Stokes wavelengths, respectively; h is Planck's constant, and c is the speed of light. The Raman frequency shift is typically 13.2 THz for optical fibers used in communication; k is the Boltzmann constant, and T is the absolute temperature of the optical cable.

[0039] Temperature calculation process: Wavelet threshold filtering is performed on the acquired Raman scattering signal, using the db4 wavelet basis to remove electromagnetic interference and junction reflection noise; Stokes / anti-Stokes light intensity peaks at the same spatial location are extracted. , Substitute the values ​​into the above formula to solve for the absolute temperature T of the optical cable, and then calculate the temperature deviation. , The reference temperature is the historical average temperature of the optical cable deployment environment, or it can be obtained through a strain-free calibration point.

[0040] Step 2022: Combine the temperature deviation and calculate the pure strain frequency shift component based on the detected Brillouin scattering signal. Combine the pure strain frequency shift component and the compensation formula to obtain the true strain parameter. First, the BOTDA / BOTDR module of the optical cable monitoring equipment detects the Brillouin scattering signal and then calculates the original value of the Brillouin frequency shift at the same spatial location. The frequency range can be 10GHz-12GHz; Substitute the known fiber temperature coefficient Calculate the frequency shift component caused by temperature. ;in, Temperature deviation; The frequency shift component corresponding to pure strain is extracted by interpolation. ; The acquired frequency shift components With fiber strain coefficient Substitute into the compensation formula The actual strain calculation is completed; among which, the fiber strain coefficient Cε is an inherent parameter of the fiber, provided by the fiber manufacturer (such as G.652 fiber). (≈0.05MHz / με), or a calibration experiment in a laboratory environment using a precision tensile device to apply standard strain and record the corresponding Brillouin frequency shift to ensure parameter accuracy.

[0041] For example, when =2MHz When =0.05MHz / με, =2MHz / 0.05MHz / με=40με, which is the actual deformation of the optical cable at this monitoring point after eliminating temperature interference, and meets the requirement of strain measurement accuracy ±5με.

[0042] Step 2023: Calibrate and verify the temperature deviation and the actual strain parameters; In the static calibration process, the sensing fiber is placed in a constant temperature chamber in a laboratory environment, and a standard strain is applied by a precision stretching device. The Brillouin frequency shift-Raman temperature data under different temperature-strain combinations are recorded, and a compensation accuracy error table is established to ensure that the strain measurement error is ≤ ±5με.

[0043] In the dynamic verification phase, after the actual optical cable deployment, a "quiet section" free from external interference is selected as the calibration point. The strain data after Raman compensation is compared in real time with the measurement results of manually attached strain gauges. Automatic calibration is performed every 24 hours to correct for strain caused by fiber aging. , drift.

[0044] Step 3: Analyze the vibration displacement data ΔL and actual strain separately. The optical cable temperature is filtered and denoised, normalized, and spatiotemporally aligned. The vibration displacement ΔL and the true strain are then processed. ε is bound to the acquired geographic coordinates; Specifically, the steps include the following: Step 301: Perform anti-interference processing on the scattered signal; In this embodiment, to address the issue of superposition of Rayleigh scattering and Brillouin scattering signals in long-distance (e.g., greater than 50km) monitoring, a combination of wavelength division multiplexing and narrowband filtering is adopted. Rayleigh scattering (vibration demodulation) and Brillouin scattering (strain / temperature demodulation) signals are separated by a 1550nm narrowband filter (bandwidth ≤ 0.5nm) to avoid crosstalk. To address the issue of Raman scattering being interfered with by joint reflection, a "reflection suppression mark" is preset at the joint location, and an algorithm is used to remove abnormal Raman signal values ​​within ±3m of the joint.

[0045] When the monitoring distance is greater than the set value, such as 80km, one fiber amplifier (EDFA) is deployed every 50km along the optical cable to compensate for the attenuation of the scattered signal and ensure that the signal-to-noise ratio of the Rayleigh scattering signal is ≥20dB and the signal-to-noise ratio of the Brillouin scattering signal is ≥18dB at the far end (such as 100km), thus ensuring demodulation accuracy.

[0046] Step 302: Obtain the location information of key monitoring nodes; Location information acquisition combines the BeiDou positioning system to obtain the precise geographic coordinates of monitoring equipment and key nodes, including poles, manholes, and junction boxes. The signal round-trip time is calculated using optical time-domain reflectometry (OTDR), with the following formula: , in, The round-trip time of the optical signal in the optical fiber is measured using optical time-domain reflectometry, which is the time interval between the emission of the optical pulse and the reception of the backscattered light. L The length of the optical fiber. To determine the speed of light propagation in optical fibers, a mapping relationship between "optical fiber length and geographic coordinates" is established, generating a spatial distribution topology map of optical cables, including attributes such as optical cable segment ID, connector location, and laying method.

[0047] Step 303: Perform noise suppression and normalization on the vibration displacement ΔL data; In the filtering and denoising stage, in addition to processing the original scattering spectrum data, noise suppression needs to be performed separately on the calculated vibration displacement ΔL data. In this embodiment, wavelet threshold filtering based on db4 wavelet basis is used (the threshold is set to 1.5 times the standard deviation of the signal) to remove the influence of electromagnetic interference (such as 50Hz power frequency interference generated by surrounding power equipment) and joint reflection noise on ΔL, so as to ensure the accuracy of vibration displacement data. For example, before filtering, the fluctuation range of ΔL is 250-270nm (including noise), and after filtering, the fluctuation range converges to 258-262nm, which is closer to the actual vibration situation.

[0048] In the normalization process, the calculated vibration displacement ΔL is included in the standardization range and processed according to the formula. = The process involves processing data where x is the original value of ΔL at a certain monitoring point, μ is the mean value of ΔL at all monitoring points in the optical cable segment, and σ is the standard deviation of ΔL. This process eliminates the dimensional deviation of ΔL caused by material differences (different refractive indices of the n optical fibers) in different optical cable segments, enabling vibration data to be analyzed in conjunction with strain and temperature data.

[0049] Step 304, and then the vibration displacement ΔL and the true strain ε is bound to the geographic coordinates obtained by the BeiDou positioning system; In the spatiotemporal alignment phase, using BeiDou time as the reference, time synchronization is performed on data with different parameters, including: Real response The ε data was timestamped and calibrated with vibration displacement ΔL, temperature T, and geographic coordinate data (error ≤ 10ms); based on the optical cable spatial topology map (including the optical fiber length-geographic coordinate mapping relationship), the monitoring points were... ε data is bound to the corresponding longitude, latitude, and optical cable segment ID (e.g., "XX optical cable segment K1+200m"). ε=35με”, to achieve spatial alignment. Simultaneously, for Outlier removal is performed on the ε data (e.g., invalid data exceeding 1.2% of the yield strain of the optical fiber material is removed) to provide accurate strain parameters for subsequent feature extraction and event localization.

[0050] Furthermore, it also includes a dynamic control mechanism for the sampling frequency, where the fluctuation amplitude of the vibration displacement ΔL exceeds a first threshold or the actual strain. When the rate of change of ε is greater than the second threshold, it is determined to be a sudden event and the first sampling frequency is triggered; otherwise, it is determined to be a slow deterioration event and a second sampling frequency lower than the first sampling frequency is activated.

[0051] Step 4: Based on the preprocessed vibration displacement ΔL and true strain ε and temperature T data are used to extract time-domain, frequency-domain, and time-frequency-domain features to form a multi-dimensional feature vector; the feature vector is then input into a pre-trained event recognition model to output the abnormal event classification result. Specifically, the steps include the following: Step 401: Based on the preprocessed vibration displacement ΔL and actual strain ε and temperature T data are used to extract time-domain, frequency-domain, and time-frequency-domain features to form a multi-dimensional feature vector. Time-domain features, including peak amplitude A_max of vibration signal and vibration energy E, are extracted based on vibration displacement ΔL. Among them, the peak amplitude of the vibration signal The calculation process is as follows: Substitute the sequence x(t) of ΔL changing with time (t is the timestamp, in seconds) into the formula. Where x(t) is the vibration displacement ΔL at a certain moment calculated by the formula; for example, if ΔL at a certain monitoring point changes with time to 260nm, 275nm, and 262nm, then... =275nm, corresponding to the peak amplitude of vibration.

[0052] The calculation process for vibration energy E is as follows: Vibration energy is calculated based on ΔL, and the formula is supplemented as follows: t1 and t2 are the start and end times of the event, The vibration displacement ΔL at a certain monitoring point at time t is represented by the integral result, which reflects the total energy of the vibration event. For example, in a mechanical excavation event, if ΔL fluctuates greatly at a certain moment, the integral result E>0.8 (standardized value) can be used as one of the criteria for judging abnormal events. This energy value is directly derived from the ΔL data calculated by the formula.

[0053] Based on real strain The temporal features, including strain gradient, are extracted from the difference in fiber length ΔL between ε and the corresponding monitoring point. and the rate of change of strain; Among them, strain gradient Calculation: Real Strain Substituting ε and the fiber length difference ΔL at the corresponding monitoring point into the formula Based on the principle of optical time-domain reflectometry, the formula is L=vt / 2, where ΔL is the difference in fiber optic length between two adjacent monitoring points, with a spatial resolution of 1m-5m. For example, adjacent monitoring point A ( ε1=30με, L1=1000m) and monitoring point B ( ε2=50με, L2=1005m), then ΔL=5m, =(50-30)με / 5m=4με / m, this value reflects the spatial rate of strain change along the optical cable, and can be used to identify events that cause strain concentration, such as soil settlement and optical cable compression (e.g. A value >8με / m indicates a high strain gradient risk.

[0054] Strain change rate calculation: based on real strain in continuous time series ε data, calculate the strain rate of change Rε=Δ( ε) / Δt (Δt is the time interval, such as 1 second), for example, at time t1 ε=30με, time t2 (t2-t1=1s) If ε = 38με, then Rε = 8με / s. This value reflects the trend of strain change over time and can be used to determine whether strain has an accumulative trend (e.g., if R_ε is consistently positive and greater than 2με / s, it indicates a risk of strain accumulation).

[0055] Frequency domain features are obtained by converting the vibration signal to the frequency domain using Fast Fourier Transform (FFT) to extract the dominant frequency, spectral peak value, and frequency band energy percentage, such as the energy percentage in the 10-50Hz band, to distinguish between excavation and traffic vibrations. The formula for the dominant frequency is: The frequency domain characteristics are generated based on the vibration displacement ΔL data through Fast Fourier Transform, and are used to distinguish the frequency characteristics of different types of vibration events.

[0056] Time-frequency domain features employ short-time Fourier transform or wavelet transform to acquire the energy distribution and abrupt change locations of the signal across different time-frequency windows. For example, a sudden temperature rise caused by a fire corresponds to a time-frequency domain energy concentration area, capturing the dynamic evolution characteristics of the event. Combining time-frequency domain features with the temporal variation of vibration displacement ΔL and abrupt temperature changes T, the parameter correlation patterns during the dynamic evolution of the event are captured.

[0057] The above is based on The time-domain characteristics of ε calculation need to be integrated with vibration and temperature characteristics to form a 44-dimensional feature vector (12-dimensional time-domain characteristics: vibration displacement ΔL, true strain). ε) + 8-dimensional frequency domain features (obtained by FFT for vibration displacement ΔL) + 24-dimensional time-frequency domain features (obtained by short-time Fourier transform / wavelet transform: energy distribution of vibration displacement ΔL in different time-frequency windows, location of abrupt change points, combined with abrupt change data of temperature T, to capture parameter correlation patterns)), all features are traced back to the core physical parameters of demodulation, providing accurate input for pattern recognition.

[0058] Step 402: Input the feature vector into the pre-trained event recognition model and output the abnormal event classification result; In the dataset construction phase, samples of six typical events were collected, including normal traffic, mechanical excavation, fiber optic cable crushing, water immersion, fire, and soil subsidence. Each event category had no fewer than 300 samples in each scenario, with a total sample size of no less than 5,400. The event type, feature parameters, and occurrence scenario were labeled, and the dataset was divided into training set, validation set, and test set in a 7:2:1 ratio.

[0059] The system covers three core fiber optic cable deployment scenarios: urban main road fiber optic cables (40%), mountainous aerial fiber optic cables (30%), and inter-regional backbone fiber optic cables (30%). For each scenario, ≥300 samples are collected for each event type (6 types in total), for a total sample size of ≥1800 samples (6 types × 3 scenarios × 100 basic samples + 600 augmented samples). Sample augmentation: Scene-specific noise (e.g., 50Hz power frequency noise superimposed on urban scenes, wind vibration noise superimposed on mountainous scenes) is added to improve the model's generalization ability. Before training, a validation set is divided by scenario (each scenario accounting for 20% of the samples in that scenario) to ensure that the event recognition accuracy in each scenario is ≥95%; otherwise, additional samples are added for the corresponding scenario.

[0060] In the model construction phase, a CNN-LSTM hybrid architecture is adopted. The input layer consists of a 44-dimensional feature vector, including 12-dimensional manually extracted features and 32-dimensional DBN automatically extracted features. The CNN layer has two convolutional layers and one max pooling layer to capture local feature correlations. The LSTM layer has one hidden layer with a dimension of 64 to capture temporal evolution patterns. The output layer is a Softmax classifier, and the model is trained using the Adam optimizer and cross-entropy loss function with a learning rate of 0.001. The cross-entropy loss function includes L2 regularization.

[0061] In the event recognition stage, real-time feature vectors are input into the trained model, with a test set accuracy of no less than 97% and a high-risk event recall rate of no less than 95%, thus achieving automatic classification of abnormal events.

[0062] The 44-dimensional feature vectors are divided into three categories based on their physical meaning and correlation with events, and then input into different layers of the CNN-LSTM hybrid architecture for targeted processing. Vibration feature set (16 dimensions): Includes peak amplitude, vibration energy, dominant frequency, and frequency band energy ratio calculated based on vibration displacement ΔL, directly input into the first convolutional layer of the CNN. This layer uses a 3×1 convolutional kernel (stride 1, padding=1), and is processed using the formula... The local time-frequency domain correlation of vibration features is extracted (where W1 is the 16×32×3 convolution kernel parameter, B1 is the bias term, and x is the vibration feature input matrix), focusing on capturing high-frequency vibration features of events such as mechanical excavation and vehicle crushing.

[0063] Strain-temperature feature set (18 dimensions): including those based on real strain The strain gradient, strain rate of change, mean temperature T, and sudden temperature rise magnitude calculated by ε are input into the second convolutional layer of the CNN. This layer uses a 5×1 convolutional kernel (stride 1, padding=2), and is processed using the formula... Further extract the synergistic variation features of strain and temperature (W2 is the 32×64×5 convolution kernel parameter, B2 is the bias term, and y is the strain-temperature feature input matrix) to identify events such as soil subsidence (abnormal strain gradient) and fire (sudden temperature rise + abrupt strain change).

[0064] The spatiotemporal correlation feature group (10-dimensional) includes geographic coordinates (longitude / latitude), feature acquisition timestamps, and fiber optic cable segment ID mapping values. After One-Hot encoding (e.g., fiber optic cable segment ID is encoded as an 8-dimensional vector according to region classification), it is concatenated with the 64-dimensional feature map output by the CNN and input together into the LSTM layer. The hidden state of the final output of the LSTM layer is then processed. (64-dimensional) Input to the Softmax classifier, using the formula Calculate the probability distribution of 6 types of events. It is a 64×6 weight matrix. (where y is the bias term and y is the event type label).

[0065] An attention mechanism is introduced to assign weights to the concatenated feature vectors, using the following formula: , Where: Q (query vector) is the input feature of the LSTM layer (74-dimensional), K (key vector) is the feature importance label matrix (based on historical event sample annotation, such as the weight of "vibration main frequency 10-50Hz" in mechanical excavation events is set to 0.8), and V (value vector) is the original concatenated feature; For the feature dimension (74), this formula strengthens the contribution of key features to event recognition (e.g., the weight of strain gradient in soil subsidence events is increased by 30%), and weakens irrelevant features (e.g., the weight of small temperature fluctuations in normal traffic scenarios is reduced by 50%). Based on over 10,000 historical event data points, the "information gain" algorithm is used to calculate feature importance. For example, in mechanical excavation events, the information gain value for "vibration frequency 10-50Hz" is 0.72 (the highest), so its weight is set to 0.8; the information gain value for "temperature fluctuation" is 0.15, so its weight is set to 0.1. The weight update cycle is every 3 months: the information gain is recalculated based on new event data, and the weights are adjusted to ensure the model adapts to changes in the scenario.

[0066] Introducing a feature matching verification mechanism: If the probability of a certain event... To further verify whether the core features corresponding to the event meet the preset thresholds, such as "mechanical excavation," which requires vibration energy E>0.8, dominant frequency 10-50Hz, and strain change rate <2με / s simultaneously; if the core feature matching degree is ≥0.9, the event classification result is confirmed; if the matching degree is <0.7, a feature backtracking mechanism is triggered, introducing a feature matching verification mechanism: if the probability of a certain event... The classification results were iteratively optimized by readjusting the CNN convolution kernel parameters and the LSTM temporal window length.

[0067] Step 403: Based on the output of the event recognition model and the preset dynamic threshold, determine and trigger abnormal event warnings; combine the optical time domain reflection principle and the optical cable spatial distribution topology map to locate the event to a specific geographical location; generate alarm information containing event type, location, and severity level; Based on a trained recognition model, abnormal events are detected in real time using dynamic thresholds. The dynamic thresholds are adaptively adjusted according to the fiber optic cable laying scenario; for example, the vibration threshold for urban roads is higher than that for farmland. When the model outputs an event probability of not less than 0.8 and the feature parameters exceed the preset threshold, an abnormal event warning is triggered, such as vibration energy E>0.8 or strain ε>800uε.

[0068] In this embodiment, the fiber optic location calculation utilizes the spatial resolution of distributed fiber optic sensing technology, with a spatial resolution of 1-5 meters, and combines this with the optical time-domain reflectometry principle to calculate the fiber optic length corresponding to the abnormal event (formula: Geographic coordinate mapping converts fiber optic cable length into precise geographic coordinates, including longitude, latitude, and geographic location name, through a spatial distribution topology map of optical cables, with a positioning error of no more than 5 meters.

[0069] In this embodiment, a structured alarm record is generated, including event type, occurrence time, fiber optic location, geographic coordinates, severity level, characteristic parameters, and handling suggestions. The severity level is set according to characteristic parameters and is divided into blue (low risk), yellow (medium risk), and red (high risk). Characteristic parameters include vibration spectrum, strain value, temperature curve, etc.

[0070] Step 5: Based on the actual strain in historical monitoring data ε data is used to calculate the health score of the optical cable; and based on historical data, a time-series prediction model is used to predict future health trends and actual strain. The changing trend of ε triggers a preventative maintenance warning when the prediction result meets preset conditions; Based on real strain from historical monitoring data ε data is used to calculate the health score of the optical cable, including: A health assessment model for optical cables is constructed based on long-term accumulated monitoring data, which requires at least three months of real-time strain data. The ε data is used to calculate a health score using a formula, ranging from 0 to 100. The formula is: , Where H is , , Here, Ne represents the weighting coefficient, D represents the number of abnormal events, and To represents the actual response time. ε is the cumulative duration exceeding the preset threshold, Tt is the total monitoring duration, and Ib is the business impact coefficient.

[0071] Determination of strain exceeding the limit duration: based on actual strain Based on ε, a strain exceedance threshold is set (determined according to the fiber material properties, such as a G.652 fiber yield strain ≈ 1.2%, with a warning threshold of 800 με and an exceedance threshold of 1000 με); statistical monitoring is conducted within the monitoring period. The cumulative duration of ε ≥ the threshold exceeding the standard is the strain exceeding the standard duration (e.g., within one month for a certain optical cable section). If the cumulative duration of strain exceeding the limit is 6 hours and the total monitoring time is 720 hours, then the duration of strain exceeding the limit / total monitoring time = 6 / 720 ≈ 0.0083.

[0072] Correlation of Health Score: Real-world Response Long-term changes in ε directly affect the proportion of "strain exceeding the standard duration", which in turn affects the health score. For example, if the proportion of strain exceeding the standard duration for a certain optical cable section is 0.05, the number of abnormal events / monitoring days is 0.1, and the business impact coefficient is 0.2, then the health score = 100 - (0.4 × 0.1 + 0.3 × 0.05 + 0.3 × 0.2) = 100 - (0.04 + 0.015 + 0.06) = 99.885 points, which is judged as "healthy". If the proportion of strain exceeding the standard duration rises to 0.3, and other parameters remain unchanged, then the health score = 100 - (0.04 + 0.09 + 0.06) = 99.81 points. Although it is still "healthy", the cumulative strain trend needs to be closely monitored.

[0073] Based on historical data, a time-series prediction model is used to predict future health trends and actual responses. The changing trend of ε includes: An LSTM time series forecasting model is used, with inputs including health data for the past 30 days, meteorological data, and actual strain. The historical time-series data of ε (stored at 1 hour per data point, totaling 720 data points) is output as the health trend and actual response over the next 30 days. The predicted trend of ε is analyzed, with a focus on the impact of strain accumulation on optical cable aging. Strain trend prediction: based on Using historical data, predict the next 30 days using an LSTM model. The maximum value, average value, and cumulative change of ε (e.g., predicting day 15). The maximum value of ε reached 950με, and on the 30th day it reached 1050με, which is close to the 1000με threshold of G.652 optical fiber (exceeding the standard threshold).

[0074] Preventative maintenance warning trigger: A "preventative maintenance warning" will be issued immediately if any of the following conditions are met: 1) The predicted health score is below 60; 2) The predicted actual strain... The maximum value of ε is close to or exceeds 90% of the yield strength of the optical fiber material (e.g., the yield strain of G.652 optical fiber is ≈1.2%, i.e., 1200με; setting the 90% threshold to 1080με, if predicted...). ε≥1080με); thirdly, prediction The daily cumulative increase in ε consistently exceeds 50 με (indicating rapid strain accumulation). The predicted value must be clearly indicated in the warning information. ε peak value, corresponding time and optical cable segment location (based on) (ε-bound geographic coordinates), it is recommended to conduct on-site inspections or replace the fiber optic cable section in advance (e.g., "predicting the location of the XX fiber optic cable section K3+500m on the 20th day"). (ε reaches 1100με, close to the yield strength threshold; it is recommended that the inspection and evaluation of this section of optical cable be completed within 5 days.) Example 2 This embodiment provides a real-time online monitoring system for the status of communication optical cables, including: The signal acquisition and demodulation module is configured to transmit probe light pulses to the optical fiber of the communication optical cable, which serves as the sensing medium, and to receive and demodulate its backscattered light signals. The data demodulation module is configured to calculate the vibration displacement of the optical cable based on the original backscattered light signal, and to calculate the actual strain and temperature of the optical cable based on the Brillouin scattering signal and Raman scattering signal obtained by separating the original backscattered light signal. The data association module is configured to perform vibration displacement, actual strain and temperature filtering and noise reduction, normalization and spatiotemporal alignment processing on the optical cable, and bind the vibration displacement and actual strain to the acquired geographic coordinates. The abnormal event classification module is configured to extract time-domain, frequency-domain, and time-frequency-domain features based on preprocessed vibration displacement, real strain, and temperature data to form a multi-dimensional feature vector. The feature vector is then input into a pre-trained event recognition model to output the abnormal event classification result. The early warning module is configured to determine and trigger early warnings for abnormal events based on the output of the event recognition model and preset dynamic thresholds.

[0075] It should be noted that the specific implementation of the real-time online monitoring system for the status of communication optical cables in this embodiment of the invention is similar to the specific implementation of the real-time online monitoring method for the status of communication optical cables in this embodiment of the invention. For details, please refer to the description in the method section. To reduce redundancy, it will not be repeated here.

[0076] Example 3 This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the real-time online monitoring method for the status of a communication optical cable as described above.

[0077] Example 4 This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the real-time online monitoring method for the status of a communication optical cable as described above.

[0078] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0079] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0080] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0081] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1The steps of the function specified in one or more boxes.

[0082] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0083] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for real-time online monitoring of the status of communication optical cables, characterized in that, Includes the following steps: The probe light pulse is transmitted into the optical fiber of the communication cable, which serves as the sensing medium, and the backscattered light signal is received and demodulated. The vibration displacement of the optical cable is calculated based on the original backscattered light signal, and the actual strain and temperature of the optical cable are calculated based on the Brillouin scattering signal and Raman scattering signal obtained by separating the original backscattered light signal. The vibration displacement, actual strain, and temperature of the optical cable are filtered and denoised, normalized, and spatiotemporally aligned respectively. The vibration displacement and actual strain are then bound to the acquired geographic coordinates. Based on the preprocessed vibration displacement, real strain and temperature data, time domain, frequency domain and time frequency domain features are extracted to form a multi-dimensional feature vector. The feature vector is then input into a pre-trained event recognition model to output the abnormal event classification result. Based on the output of the event recognition model and the preset dynamic threshold, abnormal event warnings are determined and triggered.

2. The method for real-time online monitoring of the status of a communication optical cable as described in claim 1, characterized in that, The calculation of the true strain and temperature of the optical cable based on the Brillouin scattering and Raman scattering signals obtained by separating the original backscattered light signal includes: Wavelet threshold filtering was performed on the acquired Raman scattering signals to extract the Stokes / anti-Stokes intensity peaks at the same spatial location. , The absolute temperature of the optical cable is solved by inverse calculation, and the temperature deviation is calculated. The true temperature of the optical cable is then calculated. By combining the temperature deviation, the pure strain frequency shift component is calculated based on the detected Brillouin scattering signal. The true strain parameters are then obtained by combining the pure strain frequency shift component with the compensation formula.

3. The method for real-time online monitoring of the status of a communication optical cable as described in claim 1, characterized in that, When binding vibration displacement, actual strain, and acquired geographic coordinates, the process includes timestamping the actual strain data with vibration displacement, temperature, and geographic coordinate data; establishing a "fiber optic length-geographic coordinate" mapping relationship based on the acquired location information of key monitoring nodes; generating a spatial distribution topology map of the optical cable; and binding the actual strain data of each monitoring point with the corresponding longitude, latitude, and optical cable segment ID based on the optical cable spatial distribution topology map.

4. The method for real-time online monitoring of the status of a communication optical cable as described in claim 1, characterized in that, The process involves extracting time-domain, frequency-domain, and time-frequency-domain features from preprocessed vibration displacement, actual strain, and temperature data to form a multi-dimensional feature vector. include: Time-domain features include vibration time-domain features and strain time-domain features. Vibration time-domain features include peak amplitude of vibration signal and vibration energy, while strain time-domain features include strain gradient and strain rate of change. The frequency domain features are generated based on vibration displacement data through fast Fourier transform and are used to distinguish the frequency characteristics of different types of vibration events. The time-frequency domain features are extracted by combining the temporal changes of vibration displacement with abrupt temperature changes, and by capturing the parameter correlation patterns in the dynamic evolution of events. By integrating time-domain features, frequency-domain features, and time-frequency-domain features, a multi-dimensional feature vector is constructed.

5. The method for real-time online monitoring of the status of a communication optical cable as described in claim 1, characterized in that, The event recognition model employs a CNN-LSTM hybrid architecture. During training, the constructed multi-dimensional feature vectors are classified, including vibration feature groups, strain-temperature feature groups, and spatiotemporal correlation feature groups. These are then input into different layers of the CNN-LSTM hybrid architecture. The vibration feature group is input into the first convolutional layer of the CNN to capture high-frequency vibration features, while the strain-temperature feature group is input into the second convolutional layer of the CNN to capture the coordinated changes in strain and temperature. These features are then concatenated with the spatiotemporal correlation features and input into the LSTM. The probability distribution of the event is obtained based on the hidden states output by the LSTM. The parameters of the CNN and LSTM are then adjusted according to the probability distribution of the event.

6. The method for real-time online monitoring of the status of a communication optical cable as described in claim 1, characterized in that, The method also includes calculating the health score of the optical cable based on historical real strain data; Based on historical data, a time-series prediction model is used to predict future health trends and actual strain changes. When the prediction results meet preset conditions, a preventative maintenance warning is triggered.

7. The method for real-time online monitoring of the status of a communication optical cable as described in claim 1, characterized in that, The formula for the health rating of optical cables is: , in, For health, , These are the weighting coefficients. D represents the number of abnormal events, and D represents the number of monitoring days. For realistic response The cumulative duration for which ε exceeds the preset threshold. Total monitoring duration This represents the business impact coefficient.

8. A real-time online monitoring system for the status of communication optical cables, characterized in that, include: The signal acquisition and demodulation module is configured to transmit probe light pulses to the optical fiber of the communication optical cable, which serves as the sensing medium, and to receive and demodulate its backscattered light signals. The data demodulation module is configured to calculate the vibration displacement of the optical cable based on the original backscattered light signal, and to calculate the actual strain and temperature of the optical cable based on the Brillouin scattering signal and Raman scattering signal obtained by separating the original backscattered light signal. The data association module is configured to perform vibration displacement, actual strain and temperature filtering and noise reduction, normalization and spatiotemporal alignment processing on the optical cable, and bind the vibration displacement and actual strain to the acquired geographic coordinates. The abnormal event classification module is configured to extract time-domain, frequency-domain, and time-frequency-domain features based on preprocessed vibration displacement, real strain, and temperature data to form a multi-dimensional feature vector. The feature vector is then input into a pre-trained event recognition model to output the abnormal event classification result. The early warning module is configured to determine and trigger early warnings for abnormal events based on the output of the event recognition model and preset dynamic thresholds.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the real-time online monitoring method for the status of a communication optical cable as described in any one of claims 1-7.

10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the real-time online monitoring method for the status of a communication optical cable as described in any one of claims 17.

Images