Overhead optical cable environment sensing and state monitoring method based on BOTDR

Abstract

The application discloses an overhead optical cable environment sensing and state monitoring method based on BOTDR, and relates to the technical field of computer and auxiliary equipment repair, and the method comprises the following steps: collecting optical cable strain, temperature parameters, environmental parameters and Beidou positioning data through an improved BOTDR sensor and an auxiliary sensor; carrying out denoising and format unification preprocessing on the collected data, correlating strain, temperature and environmental parameters through federated learning, generating optical cable state preliminary evaluation results and filtering invalid data; constructing an edge and cloud collaborative processing architecture, extracting fault features after receiving the preprocessed data, combining historical data to generate an optical cable health index; triggering an early warning mechanism when the optical cable health index is lower than 60 points or sudden strain occurs, integrating BOTDR and Beidou data to locate fault points, automatically dispatching work orders and pushing fault analysis reports, so that the application can realize the purposes of active early warning, accurate positioning and efficient operation and maintenance, and guarantee the stable operation of the overhead optical cable network.

Description

Technical Field

[0001] This invention belongs to the field of computer and auxiliary equipment repair technology, specifically relating to a method for environmental perception and condition monitoring of overhead optical cables based on BOTDR. Background Technology

[0002] Aerial optical cables are a critical channel for data transmission in communication networks. Their failures can directly lead to network outages, affecting the normal operation of computers and auxiliary equipment. As the core transmission carrier of communication networks, the operational status of aerial optical cables directly determines the stability and reliability of the communication network. With the popularization of new-generation information technologies such as 5G and the Internet of Things, communication traffic is growing exponentially, which places higher demands on the transmission quality and operational safety of aerial optical cables. Environmental perception and status monitoring technologies have become key supports for ensuring the efficient operation and maintenance of optical cable networks.

[0003] Currently, the environmental perception and status monitoring of overhead optical cables mainly rely on a comprehensive monitoring system based on distributed optical fiber sensing technology, combined with optical time domain reflectance (OTDR), fiber optic grating sensing, intelligent sensor networks, video surveillance, and AI. Among them, distributed optical fiber sensing technology includes Brillouin optical time domain reflectance (BOTDR), phase-sensitive optical time domain reflectance (φ-OTDR), and distributed temperature sensing (DTS), which can achieve continuous monitoring at the hundred-kilometer level and are used to measure parameters such as icing, vibration, and temperature. It has the advantages of not requiring additional power supply and having a wide monitoring range.

[0004] However, distributed optical fiber sensing technology generates terabytes of monitoring data every day. Existing data processing platforms typically use traditional cloud computing architectures, resulting in data parsing delays exceeding 10 minutes. Real-time early warning response times cannot meet the "minute-level" requirements, and the accuracy of fault prediction models is not ideal. This makes it difficult to cope with sudden fault scenarios in complex environments, affecting the stable operation of overhead optical cable networks. Summary of the Invention

[0005] The purpose of this invention is to provide a method for environmental perception and status monitoring of overhead optical cables based on BOTDR, which can achieve proactive early warning, accurate positioning and efficient operation and maintenance, and ensure the stable operation of the overhead optical cable network, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] The method for environmental sensing and condition monitoring of overhead optical cables based on BOTDR includes the following steps:

[0008] S1. Acquire optical cable strain and temperature parameters, environmental parameters, and BeiDou positioning data through an improved BOTDR sensor and auxiliary sensors;

[0009] S2. The collected data is preprocessed by denoising and format unification. By correlating strain, temperature and environmental parameters through federated learning, a preliminary assessment result of the optical cable status is generated and invalid data is filtered out.

[0010] S3. Construct an edge and cloud collaborative processing architecture, and extract fault features after receiving preprocessed data, and generate an optical cable health index by combining historical data.

[0011] The process for generating the optical cable health index is as follows:

[0012] D1. Construct an indicator system by combining the real-time characteristics of edge nodes and historical data from the cloud platform;

[0013] The indicators in the indicator system include real-time strain status, real-time temperature status, fault feature matching degree, and environmental adaptability. Among them, real-time strain status is based on the ratio of the current strain value to the threshold, real-time temperature status is based on the ratio of the current temperature value to the threshold, fault feature matching degree is the cosine similarity between the current dimensionality reduction feature and the historical fault feature, and environmental adaptability is the environmental impact degree calculated based on the association model.

[0014] D2. Calculate the score of each indicator in the indicator system;

[0015] The formula for calculating the real-time strain status index score is as follows:

[0016]

[0017] In the formula, S1 is the real-time strain state index score. This represents the strain value of the optical cable. This is the normal strain threshold. This is the abnormal strain threshold. The effective out-of-tolerance value;

[0018] optical cable strain value The formula is:

[0019]

[0020] In the formula, This is the strain coefficient, which is calibrated by the fiber material and is typically approximately 1.2 × 10⁻⁶. -6 / Hz, This represents the shift in the frequency of the scattered light due to strain.

[0021] The formula for calculating the real-time temperature status index score is as follows:

[0022]

[0023] In the formula, S2 is the score of the real-time temperature status index. This represents the temperature change of the optical cable. This is the normal temperature threshold. This is the abnormal temperature threshold. This is within the normal temperature fluctuation range;

[0024] The formula for calculating the temperature change of optical cable is:

[0025]

[0026] In the formula, This refers to the temperature coefficient, which is calibrated by the fiber material and is typically approximately 0.05 × 10⁻⁶. -6 / Hz.℃, This represents the shift in the frequency of the scattered light due to strain.

[0027] The formula for calculating the fault feature matching index score is as follows:

[0028]

[0029] In the formula, S3 is the fault feature matching degree index score. The angle between current features and historical fault features. The closer to 1, the higher the match and the lower the score;

[0030] The formula for calculating the environmental adaptability index score is: S4=100*(1-I), where S4 is the environmental adaptability index score and I is the degree of environmental impact, ranging from 0 to 1.

[0031] D3. The moving average method is used to combine the indicator scores within one week. The calculation formula is as follows:

[0032]

[0033] in, The score for the indicator within a week. The score for the i-th indicator on day j;

[0034] D4. Calculate the health index by weighting the scores of each indicator within a week. The formula for calculating the health index is:

[0035]

[0036] In the formula, The health index consists of four parts: S1 (real-time strain status score), S2 (real-time temperature status score), S3 (fault characteristic matching degree score), and S4 (environmental adaptability score). These are the weight values ​​for real-time strain status, real-time temperature status, fault characteristic matching degree, and environmental adaptability, respectively.

[0037] S4. When the optical cable health index is below 60 points or an emergency occurs, an early warning mechanism is triggered. The system integrates BOTDR and BeiDou data to locate the fault point, automatically dispatches work orders, and pushes a fault analysis report.

[0038] Preferably, the improved BOTDR sensor includes a narrow linewidth laser module, a pulse modulation module, an optical circulator module, a weak signal detection module, and a signal acquisition and processing module. The narrow linewidth laser module is connected to the pulse modulation module, the optical circulator module is connected to both the pulse modulation module and the weak signal detection module, and the signal acquisition and processing module is connected to the weak signal detection module. The narrow linewidth laser module employs a distributed feedback laser. The weak signal detection module includes an avalanche photodiode and a low-noise amplifier. The signal acquisition and processing module integrates a high-speed ADC.

[0039] Preferably, the process for acquiring optical cable strain and temperature parameters using the BOTDR sensor is as follows:

[0040] A1. The narrow linewidth laser module outputs continuous light, which is then modulated by a pulse modulation module to generate a laser pulse of a set width, which is then injected into one end of the overhead optical cable by an optical circulator.

[0041] A2. When the laser pulse is transmitted in the optical cable, Rayleigh scattering occurs due to the inhomogeneity of the fiber core. Some of the scattered light returns along the original optical path and enters the weak signal detection module through the optical circulator.

[0042] A3. The avalanche photodiode converts the scattered light signal into a current signal. After being amplified by a low-noise amplifier, the signal is sampled by a high-speed ADC to obtain the scattered light intensity data that changes over time.

[0043] A4. Parameters for calculating the strain and temperature change of optical cable based on scattered light intensity data.

[0044] Preferably, the auxiliary sensors include a positioning sensor, a temperature and humidity sensor, a wind speed and direction sensor, and an icing thickness sensor. The positioning sensor uses a Beidou dual-mode positioning module with a positioning accuracy of less than 1m for positioning. When collecting environmental parameters and Beidou positioning data, the auxiliary sensors are installed on poles along the optical cable, and the positioning sensors are fixed to the poles to ensure unobstructed satellite signal reception. The clocks of the auxiliary sensors and the BOTDR sensors are synchronized via an RS485 bus, a sampling period is set, data is collected according to the set period, and the collected data is temporarily stored.

[0045] Preferably, when denoising the collected data, a wavelet basis is selected to perform wavelet decomposition on the collected strain, temperature and environmental parameter data to obtain approximation coefficients and detail coefficients. Soft thresholding is applied to the detail coefficients, and wavelet reconstruction is performed on the processed detail coefficients and approximation coefficients to obtain the denoised data.

[0046] When processing data in a unified format, the strain and temperature data of the BOTDR, along with the temperature, humidity, wind speed, ice thickness, and BeiDou positioning data from the auxiliary sensors, are all converted into JSON format.

[0047] Preferably, when federated learning correlates strain, temperature, and environmental parameters, the following correlation process is executed:

[0048] B1. Construct a linear regression model for each monitoring point with denoised strain and temperature as dependent variables and environmental parameters as independent variables. Minimize the loss function using the gradient descent method to obtain the local model parameters.

[0049] B2. Encrypt the local model parameters using a homomorphic encryption algorithm and then upload them to the federated learning server;

[0050] B3. The federated learning server performs a weighted average of the encryption parameters of all local nodes to obtain the global correlation model parameters.

[0051] B4. The federated learning server decrypts the global correlation model parameters and distributes them to each local node. The local nodes update the model with the new data and repeat B1-B3 until the model loss function converges, thus obtaining the final correlation model of strain, temperature and environmental parameters.

[0052] Preferably, when generating the preliminary assessment results of the optical cable status, strain index, temperature index, environmental impact index, and parameter thresholds are first set, then a weighted summation is used to calculate the preliminary assessment score, and the status is divided according to the calculated score. Finally, rule filtering and statistical filtering are used to remove invalid data. When classifying the status based on the preliminary assessment score, those with a preliminary assessment score of 80 or above are classified as good, those with a preliminary assessment score between 60 and 80 are classified as average, and those with a preliminary assessment score less than 60 are classified as needing attention.

[0053] Preferably, the construction process of the edge and cloud collaborative processing architecture is as follows:

[0054] C1. Set up an edge node in each fiber optic cable coverage area, and configure an industrial computer, local database and 4G / 5G communication module to enable the edge node to receive pre-processed data;

[0055] C2. The cloud platform uses a distributed server cluster and configures a distributed database within the distributed server cluster;

[0056] C3. Edge nodes handle tasks with high real-time requirements, while the cloud platform handles non-real-time tasks. The edge nodes and the cloud platform communicate using the MQTT protocol.

[0057] Preferably, during fault feature extraction, the strain and temperature data received by the edge nodes are first sliced ​​according to a time window to obtain multiple data segments. The size of the time window is set according to the actual feature extraction requirements. Then, the statistical features, trend features, and mutation features of each data segment are calculated. Principal component analysis algorithm is used to reduce the dimensionality of the extracted multi-dimensional features to obtain the fault feature vector.

[0058] The BOTDR-based method for environmental sensing and status monitoring of overhead optical cables proposed in this invention has the following advantages compared with existing technologies:

[0059] 1. This invention uses an improved BOTDR sensor to accurately measure the strain and temperature of optical cables, combined with multiple auxiliary sensors, eliminating data blind spots; data processing employs an edge and cloud collaborative processing architecture, providing preliminary early warnings within seconds, and federated learning balances model accuracy and data privacy; early warnings utilize a dual trigger mechanism to reduce false alarms and missed alarms; fault location integrates BOTDR and BeiDou, with linear interpolation providing accurate latitude and longitude; operation and maintenance achieve automatic work order generation and intelligent dispatch, with cloud-based data storage for model optimization, forming a closed loop of monitoring, early warning, repair, and optimization, comprehensively ensuring the stability of the overhead optical cable network;

[0060] 2. This invention achieves preliminary early warning within seconds through an edge and cloud collaborative processing architecture, improving the accuracy of cloud-based fault prediction and ensuring the stability of power supply and communication in remote mountainous areas;

[0061] 3. This invention uses federated learning to construct a linear regression model of strain, temperature and environmental parameters with each monitoring point as a local node. By encrypting the upload and using global weighted averaging, a global correlation model is generated to quantify the impact of the environment on the optical cable status and identify potential risks in advance. Attached Figure Description

[0062] Figure 1 A flowchart according to an embodiment of the present invention is shown;

[0063] Figure 2 A block diagram of an improved BOTDR sensor according to an embodiment of the present invention is shown;

[0064] Figure 3 A flowchart illustrating the acquisition process of optical cable strain and temperature parameters according to an embodiment of the present invention is shown.

[0065] Figure 4 A flowchart illustrating the federated learning of strain, temperature, and environmental parameters according to an embodiment of the present invention is shown.

[0066] Figure 5 A flowchart illustrating the process of generating the optical cable health index according to an embodiment of the present invention is shown. Detailed Implementation

[0067] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0068] This invention provides, for example Figure 1-5 The BOTDR-based method for environmental sensing and status monitoring of overhead optical cables, as shown, includes the following steps:

[0069] S1. Acquire optical cable strain and temperature parameters, environmental parameters, and BeiDou positioning data through an improved BOTDR sensor and auxiliary sensors;

[0070] The improved BOTDR sensor includes a narrow linewidth laser module, a pulse modulation module, an optical circulator module, a weak signal detection module, and a signal acquisition and processing module. The narrow linewidth laser module is connected to the pulse modulation module, the optical circulator module is connected to both the pulse modulation module and the weak signal detection module, and the signal acquisition and processing module is connected to the weak signal detection module.

[0071] The narrow-linewidth laser module employs a distributed feedback laser with an output wavelength of 1550nm and a linewidth ≤10kHz, ensuring accuracy during long-distance signal transmission. The pulse modulation module converts continuous laser light into nanosecond-level pulses via an electro-optic modulator, achieving spatial resolution control. The optical circulator module isolates the transmission and reception paths, preventing direct interference from the transmitted light to the receiver and ensuring that only backscattered Rayleigh light from the optical cable is collected. The weak signal detection module includes an avalanche photodiode and a low-noise amplifier, converting weak scattered light signals into electrical signals and amplifying them to a processable level. The signal acquisition and processing module integrates a high-speed ADC for initial sampling of the electrical signals, while simultaneously suppressing high-frequency noise through digital filtering.

[0072] The process for acquiring optical cable strain and temperature parameters using the BOTDR sensor is as follows:

[0073] A1. The narrow linewidth laser module outputs continuous light, which is then modulated by a pulse modulation module to generate a laser pulse of a set width, which is then injected into one end of the overhead optical cable by an optical circulator.

[0074] A2. When the laser pulse is transmitted in the optical cable, Rayleigh scattering occurs due to the inhomogeneity of the fiber core. Some of the scattered light returns along the original optical path and enters the weak signal detection module through the optical circulator.

[0075] A3. The avalanche photodiode converts the scattered light signal into a current signal. After being amplified by a low-noise amplifier, the signal is sampled by a high-speed ADC to obtain the scattered light intensity data that changes over time.

[0076] A4. Parameters for calculating optical cable strain and temperature change based on scattered light intensity data. The formula for calculating optical cable strain is:

[0077]

[0078] In the formula, This represents the strain value of the optical cable. The strain coefficient is determined by the fiber material and is typically approximately 1.2 × 10⁻⁶. -6 / Hz, This represents the shift in the frequency of the scattered light due to strain.

[0079] The formula for calculating the temperature change of optical cable is:

[0080]

[0081] In the formula, This represents the temperature change of the optical cable. This refers to the temperature coefficient, which is calibrated by the fiber material and is typically approximately 0.05 × 10⁻⁶. -6 / Hz.℃, This represents the shift in the frequency of the scattered light due to strain.

[0082] The auxiliary sensors include a positioning sensor, a temperature and humidity sensor, a wind speed and direction sensor, and an ice thickness sensor. The temperature and humidity sensor has a measurement range of -40 to 85℃ and an accuracy of ±0.5℃. The wind speed and direction sensor has a measurement range of 0-60m / s and an accuracy of ±0.3m / s. The ice thickness sensor is based on the capacitance principle, has a measurement range of 0-100mm, and an accuracy of ±0.5mm. The positioning sensor uses a Beidou dual-mode positioning module with a positioning accuracy of less than 1m for positioning.

[0083] When collecting environmental parameters and BeiDou positioning data, auxiliary sensors are installed on poles along the optical cable, and the positioning sensors are fixed to the poles to ensure unobstructed reception of satellite signals. The clocks of the auxiliary sensors and the BOTDR sensors are synchronized via RS485 bus, a sampling period is set, data is collected according to the set period, and the collected data is temporarily stored.

[0084] S2. The collected data is preprocessed by denoising and format unification. By correlating strain, temperature and environmental parameters through federated learning, a preliminary assessment result of the optical cable status is generated and invalid data is filtered out.

[0085] When denoising the collected data, wavelet basis is selected to perform wavelet decomposition on the collected strain, temperature and environmental parameter data to obtain approximation coefficients and detail coefficients. Soft thresholding is applied to the detail coefficients, and wavelet reconstruction is performed on the processed detail coefficients and approximation coefficients to obtain the denoised data.

[0086] The formula for calculating the soft threshold is:

[0087]

[0088] in, The threshold parameter for dynamically adjusting the detail coefficients. The noise standard deviation is estimated from the decomposed detail coefficients, and N is the data length.

[0089] The formula for wavelet reconstruction is:

[0090]

[0091] in, For the denoised data, This is the inverse wavelet transform. These are the detail coefficients after processing. These are three levels of approximation coefficients;

[0092] When processing data in a unified format, the strain and temperature data from the BOTDR, along with the temperature, humidity, wind speed, ice thickness, and BeiDou positioning data from the auxiliary sensors, are all converted into JSON format. The JSON format fields include the acquisition time, parameter type, value, and monitoring point ID. Using the timestamp of the BeiDou positioning data as a reference, other parameter data are linearly interpolated to ensure that all parameters at the same monitoring point correspond one-to-one in the time dimension.

[0093] When federated learning correlates strain, temperature, and environmental parameters, the following correlation process is executed:

[0094] B1. Construct a linear regression model for each monitoring point with denoised strain and temperature as dependent variables and environmental parameters as independent variables. Minimize the loss function using the gradient descent method to obtain the local model parameters.

[0095] The formula for the linear regression model is:

[0096]

[0097] Where Y is the dependent variable consisting of the strain and temperature of the optical cable, and X is the matrix of independent variables consisting of environmental parameters and constant terms. For local model parameters, This is the error term;

[0098] The formula for minimizing the loss function is:

[0099]

[0100] in, Here, n represents the optimal local model parameters after applying the loss function, and n is the total number of samples. Let j be the dependent variable of the j-th sample. Let j be the independent variable for the j-th sample. Let j be the parameter vector to be optimized for the j-th sample;

[0101] B2. Encrypt the local model parameters using a homomorphic encryption algorithm and then upload them to the federated learning server;

[0102] The formula for homomorphic encryption algorithm is:

[0103]

[0104] in, For encrypted data, and These are the encryption functions for parameters x4 and y4, respectively. This is an addition operation on the ciphertext;

[0105] B3. The federated learning server performs a weighted average of the encryption parameters of all local nodes to obtain the global correlation model parameters.

[0106] The formula for calculating the parameters of the global association model is:

[0107]

[0108] in, Here are the parameters for the global association model, and k is the total number of local nodes. Let i be the amount of data in the i-th local node. These are the model parameters for the i-th local node;

[0109] B4. After decrypting the global correlation model parameters, the federated learning server distributes them to each local node. The local nodes update the model with the new data and repeat B1-B3 until the model loss function converges, thus obtaining the final correlation model of strain, temperature and environmental parameters.

[0110] The formula for the correlation model is as follows:

[0111]

[0112] in, The strain parameters for the final version of the correlation model. Here, RH is the temperature value, V is the humidity value, and h is the icing thickness. to These are the parameters corresponding to the global model;

[0113] By using federated learning with each monitoring point as a local node, a linear regression model of strain, temperature, and environmental parameters is constructed. Through encrypted uploading and global weighted averaging, a global correlation model is generated to quantify the impact of the environment on the optical cable status and identify potential risks in advance.

[0114] When generating the preliminary assessment results for the optical cable condition, strain, temperature, environmental impact, and parameter thresholds are first set. Then, a weighted summation is used to calculate the preliminary assessment score. The condition is classified according to the calculated score. Finally, rule-based filtering and statistical filtering are used to remove invalid data. The standard for the strain index is set as: normal ≤100μ Warning: 100-150μ Abnormalities > 150μ The temperature index is set as follows: -20 to 60℃ is normal, -30 to -20℃ or 60 to 70℃ is a warning, and below -30℃ or above 70℃ is abnormal. The environmental impact index is set by calculating the impact of environmental parameters on the optical cable using a correlation model, i.e.:

[0115]

[0116] In the formula: 25 represents normal temperature, 45℃ represents the temperature fluctuation range, 60 represents the maximum wind speed, and 100 represents the maximum icing thickness. Here, I represents temperature, V represents wind speed, and h represents icing thickness. A value of I ≤ 0.3 indicates normal conditions, I between 0.3 and 0.6 indicates a warning, and I > 0.6 indicates an abnormal condition. The weighted summation formula is:

[0117] and ,in, For the initial assessment of the score, To score the strain index, Score the temperature index. To score the environmental impact indicators, These are the weighting coefficients for strain indicators, temperature indicators, and environmental impact indicators, respectively. When classifying states based on preliminary assessment scores, scores ≥80 are classified as "Good," scores between 60-80 as "Average," and scores <60 as "Needs Attention." Rule-based filtering removes data with values ​​exceeding the sensor's range, data missing rates greater than 10%, or weak BeiDou positioning signals. Statistical filtering applies a different method to the rule-filtered data. The criteria are further filtered to eliminate those that exceed the limits. Data within the scope of the criteria are retained. Data from the guidelines;

[0118] S3. Construct an edge and cloud collaborative processing architecture, and extract fault features after receiving preprocessed data, and generate an optical cable health index by combining historical data.

[0119] The construction process of the edge and cloud collaborative processing architecture is as follows:

[0120] C1. Set up an edge node in each fiber optic cable coverage area, and configure an industrial computer, local database and 4G / 5G communication module to enable the edge node to receive pre-processed data, realize real-time fault feature extraction, short-term data storage, preliminary judgment of early warning, and upload only fault data and periodic summary data to the cloud platform.

[0121] C2. The cloud platform adopts a distributed server cluster and configures a distributed database in the distributed server cluster for long-term data storage, historical data analysis, global calculation of health index and fault model update, and sends the updated model parameters to edge nodes.

[0122] C3. Edge nodes handle tasks with high real-time requirements, such as fault feature extraction, while the cloud platform handles non-real-time tasks, such as historical data fusion. The edge nodes and the cloud platform communicate using the MQTT protocol to ensure data transmission reliability.

[0123] During fault feature extraction, the strain and temperature data received by the edge nodes are first sliced ​​according to a time window to obtain multiple data segments. The size of the time window is set according to the actual feature extraction requirements. Then, the statistical features, trend features, and abrupt change features of each data segment are calculated. Principal component analysis is used to reduce the dimensionality of the extracted multi-dimensional features to obtain the fault feature vector. Among them, the statistical features include the mean, variance, maximum value, and minimum value, and the trend feature is the linear fitting slope. The formula for calculating the linear fitting slope is:

[0124]

[0125] in, A set of data points The slope, where m is the total number of data points. and Let be the values ​​of the independent and dependent variables for the i-th data point, respectively.

[0126] The mutation feature is determined based on the difference between adjacent data. When the difference is greater than the mutation threshold, it is marked as a mutation feature. The mutation threshold is obtained by statistical analysis of historical fault data.

[0127] When principal component analysis (PCA) reduces the dimensionality of extracted multi-dimensional features, it first calculates the covariance matrix of the feature matrix, then solves for the eigenvalues ​​and corresponding eigenvectors of the covariance matrix. The top p features with a cumulative contribution rate ≥ 85% are selected, and their corresponding eigenvectors form the projection matrix U = [u1, u2, ..., up]. The dimensionality-reduced eigenvector is Y = X × U, where the formula for the covariance matrix is:

[0128]

[0129] Where C is the covariance matrix and M is the sample size. Sample mean For the sample matrix, For the transpose of the centralized data matrix, The data matrix is ​​centralized; the process for generating the optical cable health index is as follows:

[0130] D1. Construct an indicator system by combining the real-time characteristics of edge nodes and the historical data of the cloud platform. The indicators in the indicator system include real-time strain status, real-time temperature status, fault feature matching degree, and environmental adaptability. Among them, the real-time strain status is based on the ratio of the current strain value to the threshold, the real-time temperature status is based on the ratio of the current temperature value to the threshold, the fault feature matching degree is the cosine similarity between the current dimensionality reduction feature and the historical fault feature, and the environmental adaptability is the environmental impact degree calculated based on the association model.

[0131] D2. Calculate the score for each indicator in the indicator system; the formula for calculating the score of the real-time strain status indicator is:

[0132]

[0133] In the formula, S1 is the real-time strain state index score. This represents the strain value of the optical cable. This is the normal strain threshold. This is the abnormal strain threshold. For valid out-of-tolerance values, the real-time temperature status index score is calculated using the following formula:

[0134]

[0135] In the formula, S2 is the score of the real-time temperature status index. This represents the temperature change of the optical cable. This is the normal temperature threshold. This is the abnormal temperature threshold. The normal temperature fluctuation range is defined by the following formula: [Formula for calculating the fault characteristic matching score]

[0136]

[0137] In the formula, S3 is the fault feature matching degree index score. The angle between current features and historical fault features. The closer to 1, the higher the matching degree and the lower the score; the formula for calculating the environmental adaptability index score is: S4=100×(1-I), where S4 is the environmental adaptability index score and I is the environmental impact degree, ranging from 0 to 1.

[0138] D3. The moving average method is used to combine the indicator scores within one week. The calculation formula is as follows:

[0139]

[0140] in, The score for the indicator within a week. The score for the i-th indicator on day j;

[0141] D4. Calculate the health index by weighting the scores of each indicator within a week. The formula for calculating the health index is:

[0142]

[0143] In the formula, The health index consists of four parts: S1 (real-time strain status score), S2 (real-time temperature status score), S3 (fault characteristic matching degree score), and S4 (environmental adaptability score). These are the weight values ​​for real-time strain status, real-time temperature status, fault characteristic matching degree, and environmental adaptability, respectively.

[0144] S4. When the optical cable health index is below 60 points or an emergency occurs, an early warning mechanism is triggered, integrating BOTDR and Beidou data to locate the fault point, automatically dispatching work orders and pushing fault analysis reports.

[0145] The early warning mechanism classifies warnings based on specific scores and implements corresponding response measures for each level. The early warning levels include blue, yellow, and red. A blue warning occurs when the optical cable health index is between 50 and 60 points, with the response measures being: edge nodes send reminder SMS messages to the operations and maintenance center, and the health index is updated every 30 minutes. A yellow warning occurs when the optical cable health index is between 40 and 50 points or during an emergency, with the response measures being: the operations and maintenance center arranges personnel for on-site inspections, updates data every 10 minutes, and shuts down non-critical business channels on the optical cable. A red warning occurs when the optical cable health index is less than 40 points or the emergency exceeds the abnormal emergency threshold, with the response measures being: activating the emergency response team, interrupting optical cable services (switching to a backup link), and on-site repair personnel arriving within one hour.

[0146] When integrating BOTDR and BeiDou data to locate fault points, the distance between the fault point and the monitoring starting point is calculated based on the time difference of scattered light from the BOTDR. Then, BeiDou positioning data from two nearby towers is acquired, and the straight-line distance of the optical cable between the towers is calculated. If the distance between the fault point and the starting point is between the BeiDou positioning data of the two nearby towers, linear interpolation is used to calculate the coordinates of the fault point. Finally, the latitude and longitude coordinates of the fault point are output. The formula for calculating the distance between the fault point and the monitoring starting point is:

[0147]

[0148] in, The straight-line distance between the fault point and the monitoring starting point. The speed of light in a vacuum. This is the time difference between laser pulse emission and the reception of scattered light from the fault point. The refractive index of the optical fiber;

[0149]

[0150] in, This represents the straight-line distance of the optical cable between adjacent towers. and These are the BeiDou coordinates of the two towers; the linear interpolation calculation formula is:

[0151]

[0152]

[0153] Where x and y are the latitude and longitude coordinates of the final output fault point, respectively. The straight-line distance between the fault point and the monitoring starting point. and These are the BeiDou coordinates of the two towers, The standard work order is the straight-line distance between adjacent towers. When automatically dispatching work orders, the cloud platform generates a standard work order based on fault information. The fault information includes the coordinates of the fault point, the fault type, the optical cable health index, and environmental parameters. The standard work order includes the work order number, fault description, priority, and required processing time. Priorities are categorized as red (Level 1), yellow (Level 2), and blue (Level 3). The processing time is 1 hour for Level 1, 4 hours for Level 2, and 24 hours for Level 3. Based on the fault point location, the system queries the maintenance personnel database and selects the nearest personnel who are currently not assigned any tasks. The matching rule is: if there are multiple personnel, priority is given to those with ≥3 years of fault handling experience. The work order is pushed to the matched personnel through the maintenance APP, along with a fault point navigation link. If the request is not read within 5 minutes, the system automatically calls the personnel to remind them. By using an improved BOTDR sensor segmented monitoring system and BeiDou fusion positioning, the system reduces fault point positioning deviation and shortens fault investigation time. Through an edge and cloud collaborative processing architecture, it achieves second-level preliminary warnings, improves the accuracy of cloud-based fault prediction, and ensures the stability of power supply and communication in remote mountainous areas.

[0154] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for environmental sensing and condition monitoring of overhead optical cables based on BOTDR, characterized in that: Includes the following steps: S1. Acquire optical cable strain and temperature parameters, environmental parameters, and BeiDou positioning data through an improved BOTDR sensor and auxiliary sensors; S2. The collected data is preprocessed by denoising and format unification. By correlating strain, temperature and environmental parameters through federated learning, a preliminary assessment result of the optical cable status is generated and invalid data is filtered out. S3. Construct an edge and cloud collaborative processing architecture, and extract fault features after receiving preprocessed data, and generate an optical cable health index by combining historical data. The process for generating the optical cable health index is as follows: D1. Construct an indicator system by combining the real-time characteristics of edge nodes and historical data from the cloud platform; The indicators in the indicator system include real-time strain status, real-time temperature status, fault feature matching degree, and environmental adaptability. Among them, real-time strain status is based on the ratio of the current strain value to the threshold, real-time temperature status is based on the ratio of the current temperature value to the threshold, fault feature matching degree is the cosine similarity between the current dimensionality reduction feature and the historical fault feature, and environmental adaptability is the environmental impact degree calculated based on the association model. D2. Calculate the score of each indicator in the indicator system; The formula for calculating the real-time strain status index score is as follows: In the formula, S1 is the score of the real-time strain state index. This represents the strain value of the optical cable. This is the normal strain threshold. This is the abnormal strain threshold. The effective out-of-tolerance value; optical cable strain value The formula is: In the formula, This is the strain coefficient, which is calibrated by the fiber material and is typically approximately 1.2 × 10⁻⁶. -6 / Hz, This represents the shift in the frequency of the scattered light due to strain. The formula for calculating the real-time temperature status index score is as follows: In the formula, S2 is the real-time temperature status index score. This represents the temperature change of the optical cable. This is the normal temperature threshold. This is the abnormal temperature threshold. This is within the normal temperature fluctuation range; The formula for calculating the temperature change of optical cable is: In the formula, This refers to the temperature coefficient, which is calibrated by the fiber material and is typically approximately 0.05 × 10⁻⁶. -6 / Hz.℃, This represents the shift in the frequency of the scattered light due to strain. The formula for calculating the fault feature matching index score is as follows: In the formula, S3 is the fault feature matching degree index score. The angle between current features and historical fault features. The closer to 1, the higher the match and the lower the score; The formula for calculating the environmental adaptability index score is: S4=100*(1-I), where S4 is the environmental adaptability index score and I is the degree of environmental impact, ranging from 0 to 1. D3. The moving average method is used to combine the indicator scores within one week. The calculation formula is as follows: in, The score for the indicator within a week. The score for the i-th indicator on day j; D4. Calculate the health index by weighting the scores of each indicator within a week. The formula for calculating the health index is: In the formula, The health index consists of four parts: S1 (real-time strain status score), S2 (real-time temperature status score), S3 (fault characteristic matching degree score), and S4 (environmental adaptability score). These are the weight values ​​for real-time strain status, real-time temperature status, fault characteristic matching degree, and environmental adaptability, respectively. S4. When the optical cable health index is below 60 points or an emergency occurs, an early warning mechanism is triggered. The system integrates BOTDR and BeiDou data to locate the fault point, automatically dispatches work orders, and pushes a fault analysis report.

2. The method for environmental sensing and status monitoring of overhead optical cables based on BOTDR according to claim 1, characterized in that: The improved BOTDR sensor includes a narrow-linewidth laser module, a pulse modulation module, an optical circulator module, a weak signal detection module, and a signal acquisition and processing module. The narrow-linewidth laser module is connected to the pulse modulation module, the optical circulator module is connected to both the pulse modulation module and the weak signal detection module, and the signal acquisition and processing module is connected to the weak signal detection module. The narrow-linewidth laser module employs a distributed feedback laser. The weak signal detection module includes an avalanche photodiode and a low-noise amplifier. The signal acquisition and processing module integrates a high-speed ADC.

3. The method for environmental perception and status monitoring of overhead optical cables based on BOTDR according to claim 2, characterized in that: The process for acquiring optical cable strain and temperature parameters using the BOTDR sensor is as follows: A1. The narrow linewidth laser module outputs continuous light, which is then modulated by a pulse modulation module to generate a laser pulse of a set width, which is then injected into one end of the overhead optical cable by an optical circulator. A2. When the laser pulse is transmitted in the optical cable, Rayleigh scattering occurs due to the inhomogeneity of the fiber core. Some of the scattered light returns along the original optical path and enters the weak signal detection module through the optical circulator. A3. The avalanche photodiode converts the scattered light signal into a current signal. After being amplified by a low-noise amplifier, the signal is sampled by a high-speed ADC to obtain the scattered light intensity data that changes over time. A4. Parameters for calculating the strain and temperature change of optical cable based on scattered light intensity data.

4. The method for environmental perception and status monitoring of overhead optical cables based on BOTDR according to claim 3, characterized in that: The auxiliary sensors include a positioning sensor, a temperature and humidity sensor, a wind speed and direction sensor, and an icing thickness sensor. The positioning sensor uses a Beidou dual-mode positioning module with a positioning accuracy of less than 1m for positioning. When collecting environmental parameters and Beidou positioning data, the auxiliary sensors are installed on poles along the optical cable, and the positioning sensors are fixed to the poles to ensure unobstructed satellite signal reception. The clocks of the auxiliary sensors and the BOTDR sensors are synchronized via an RS485 bus, a sampling period is set, data is collected according to the set period, and the collected data is temporarily stored.

5. The method for environmental perception and status monitoring of overhead optical cables based on BOTDR according to claim 4, characterized in that: When denoising the collected data, wavelet basis is selected to perform wavelet decomposition on the collected strain, temperature and environmental parameter data to obtain approximation coefficients and detail coefficients. Soft thresholding is applied to the detail coefficients, and wavelet reconstruction is performed on the processed detail coefficients and approximation coefficients to obtain the denoised data. When processing data in a unified format, the strain and temperature data of the BOTDR, along with the temperature, humidity, wind speed, ice thickness, and BeiDou positioning data from the auxiliary sensors, are all converted into JSON format.

6. The method for environmental perception and status monitoring of overhead optical cables based on BOTDR according to claim 5, characterized in that: When federated learning correlates strain, temperature, and environmental parameters, the following correlation process is executed: B1. Construct a linear regression model for each monitoring point with denoised strain and temperature as dependent variables and environmental parameters as independent variables. Minimize the loss function using the gradient descent method to obtain the local model parameters. B2. Encrypt the local model parameters using a homomorphic encryption algorithm and then upload them to the federated learning server; B3. The federated learning server performs a weighted average of the encryption parameters of all local nodes to obtain the global correlation model parameters. B4. The federated learning server decrypts the global correlation model parameters and distributes them to each local node. The local nodes update the model with the new data and repeat B1-B3 until the model loss function converges, thus obtaining the final correlation model of strain, temperature and environmental parameters.

7. The method for environmental perception and status monitoring of overhead optical cables based on BOTDR according to claim 6, characterized in that: When generating the preliminary assessment results of the optical cable condition, strain index, temperature index, environmental impact index and parameter thresholds are first set. Then, a weighted summation is used to calculate the preliminary assessment score. The condition is classified according to the calculated score. Finally, rule filtering and statistical filtering are used to remove invalid data. When classifying the condition based on the preliminary assessment score, those with a preliminary assessment score of 80 or above are classified as good, those with a preliminary assessment score between 60 and 80 are classified as average, and those with a preliminary assessment score less than 60 are classified as needing attention.

8. The method for environmental perception and status monitoring of overhead optical cables based on BOTDR according to claim 7, characterized in that: The construction process of the edge and cloud collaborative processing architecture is as follows: C1. Set up an edge node in each fiber optic cable coverage area, and configure an industrial computer, local database and 4G / 5G communication module to enable the edge node to receive pre-processed data; C2. The cloud platform uses a distributed server cluster and configures a distributed database within the distributed server cluster; C3. Edge nodes handle tasks with high real-time requirements, while the cloud platform handles non-real-time tasks. The edge nodes and the cloud platform communicate using the MQTT protocol.

9. The method for environmental perception and status monitoring of overhead optical cables based on BOTDR according to claim 8, characterized in that: When extracting fault features, the strain and temperature data received by the edge nodes are first sliced ​​according to the time window to obtain multiple data segments. The size of the time window is set according to the actual feature extraction requirements. Then, the statistical features, trend features and mutation features of each data segment are calculated. The principal component analysis algorithm is used to reduce the dimensionality of the extracted multi-dimensional features to obtain the fault feature vector.

Images