Millimeter wave optical fiber communication transmission fault detection method and system and storage medium

Abstract

The invention discloses a millimeter wave optical fiber communication transmission fault detection method, and relates to the technical field of millimeter wave communication, and the method comprises the steps: respectively installing signal couplers at the input end and the output end of a filter of a millimeter wave optical fiber communication link, generating an original reference signal and a signal to be detected, and outputting a frequency mixing signal after coherent superposition and balance detection; carrying out trend removal and noise filtering on the frequency mixing signal to obtain a chaos characteristic signal, calculating time delay and embedded dimensions through phase space reconstruction, constructing a phase space matrix, and judging a system chaos state based on a maximum Lyapunov index; the correlation dimension and the Kolmogorov entropy of the chaotic feature signal are calculated, a feature matrix is constructed, and principal component analysis is carried out to obtain an optimized feature set; a mapping relation library of time delay fluctuation and an optimized feature set is established, a multi-level fault threshold value is set, the fault level is judged according to a comparison result of the current optimized feature set and the fault threshold value, and the accuracy and the real-time performance of fault detection are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of millimeter-wave communication technology, specifically to a method, system, and storage medium for detecting faults in millimeter-wave optical fiber communication transmission. Background Technology

[0002] Millimeter-wave optical fiber communication technology has been widely used in 5G communication, satellite communication and other fields due to its advantages such as high bandwidth and low latency. In existing technologies, time-domain reflectometry, spectrum analysis or deep learning-based fault detection methods are usually used to monitor communication links. Time-domain reflectometry locates the fault point by analyzing the time delay of the reflected signal, while spectrum analysis identifies anomalies by detecting changes in the frequency domain characteristics of the signal. Some studies use deep learning models to achieve automatic classification of fault modes by training a large amount of data.

[0003] However, existing technologies have significant limitations. Time-domain reflectometry and spectral analysis are insensitive to sub-nanosecond delay fluctuations or tiny phase jumps, making it difficult to accurately detect chaotic faults caused by filter delay fluctuations. These methods typically rely on preset thresholds or fixed patterns, failing to dynamically adapt to complex and ever-changing communication environments, resulting in insufficient accuracy and real-time performance in fault detection. Furthermore, while deep learning-based methods possess some adaptive capabilities, they require a large amount of labeled data for training, and the models have poor interpretability. In practical applications, facing complex fault modes caused by factors such as temperature variations and device aging in millimeter-wave fiber optic communication links, existing methods often struggle to balance high accuracy and low false alarm rates, failing to meet the requirements of high-reliability communication systems. Summary of the Invention

[0004] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a method, system, and storage medium for detecting faults in millimeter-wave optical fiber communication transmission. By combining chaotic signal analysis with principal component dimensionality reduction technology, it dynamically detects chaotic faults caused by filter delay fluctuations. This solves the problems of existing methods being insensitive to sub-nanosecond delay fluctuations, having poor adaptability, and relying on large amounts of data for deep learning models, thus significantly improving the accuracy and real-time performance of fault detection.

[0005] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: a method for detecting faults in millimeter-wave optical fiber communication transmission, comprising: Signal couplers are installed at the input and output ends of the filter in the millimeter-wave optical fiber communication link to generate the original reference signal and the signal to be detected, which are then modulated onto the optical carrier. After coherent superposition and balanced detection, the mixed signal is output. The mixed signal is detrended and noise filtered to obtain chaotic feature signals. The time delay and embedding dimension are calculated through phase space reconstruction to construct the phase space matrix. The chaotic state of the system is determined based on the maximum Lyapunov exponent. Calculate the correlation dimension and Kolmogorov entropy of the chaotic feature signals, construct the feature matrix, and perform principal component analysis to obtain the optimized feature set; Establish a mapping relationship library between latency fluctuations and optimized feature sets, set multi-level fault thresholds, and determine the fault level based on the comparison results between the current optimized feature set and the fault thresholds.

[0006] Furthermore, the signal couplers installed at the input and output ends of the filter are, in sequence, a first signal coupler and a second signal coupler. The clock synchronization module provides a synchronous clock for the two signal acquisition channels, and the delay compensation unit completes the synchronization calibration of the two acquired signals. The amplitude of the two acquired signals after synchronization calibration is normalized, and high-frequency noise is filtered out by a low-pass filter to obtain the original reference signal and the signal to be detected.

[0007] Furthermore, a continuous optical carrier is generated by a narrow-linewidth laser. The optical carrier is input into a 1×2 fiber beam splitter and split into two sub-optical carriers. The original reference signal and the signal to be detected are loaded onto the two sub-optical carriers by external modulation, resulting in two modulated optical signals. The two modulated optical signals are coherently superimposed by a fiber coupler with a splitting ratio of 1:1, converted into electrical signals, and then processed by a low-noise amplifier and an adaptive equalizer to output a mixed signal.

[0008] Furthermore, the least squares method is used to detrend the mixing signal to eliminate the linear drift component. Wavelet thresholding is then used to filter out noise from the detrended mixing signal, resulting in a clean chaotic feature signal. The time delay and embedding dimension required for phase space reconstruction are calculated. The time delay is determined by calculating the autocorrelation function of the chaotic feature signal, and the time value corresponding to the first drop of the autocorrelation function to 1 / e of its maximum value is taken as the time delay. The embedding dimension is determined by the spurious neighbor method, gradually increasing the embedding dimension while calculating the proportion of spurious neighbors in each dimension. When the proportion of spurious neighbors first drops below 5%, the corresponding dimension is the final embedding dimension. Based on the determined time delay and embedding dimension, the phase space matrix is ​​constructed using the delayed coordinate method.

[0009] Furthermore, by randomly selecting an initial reference point from the phase space matrix, a nearest neighbor point that does not coincide with the initial reference point is found within a small distance neighborhood of the initial reference point. The Euclidean distance between the initial reference point and the nearest neighbor point is calculated and denoted as the initial distance. The evolution trajectories of the initial reference point and the nearest neighbor point over time are tracked, and the Euclidean distance between the two corresponding evolution points at different evolution times is calculated sequentially. The natural logarithm of the distance values ​​at all evolution times is taken, and a linear fit is performed with the number of evolution steps as the abscissa and the corresponding natural logarithm value as the ordinate. The slope of the fitted line is the maximum Lyapunov exponent. When the maximum Lyapunov exponent is greater than 0, it is determined that the system corresponding to the chaotic feature signal has a chaotic phenomenon.

[0010] Furthermore, within a defined scale range, a correlation integral is constructed, and the correlation dimension is obtained through linear fitting. Based on the time delay, the mutual information value of the signal at each delay time is calculated, and the Kolmogorov entropy is obtained by fitting the decay rate of the mutual information value. The maximum Lyapunov exponent, correlation dimension, and Kolmogorov entropy are organized into a feature matrix, and after standardization, principal component analysis is performed. Principal components with a cumulative variance contribution rate of 85% or higher are selected to form an optimized feature set. Principal components with an absolute value of correlation coefficient with fault severity less than a correlation threshold are removed, and the fault discrimination capability of the optimized feature set is verified through cross-validation until the preset accuracy is met.

[0011] Furthermore, the optimized feature sets under different combinations of time delay fluctuation gradients and operating temperatures are obtained, and a mapping relationship library between time delay fluctuations and optimized feature sets is formed; the warning threshold, minor fault threshold range and severe fault threshold are calibrated to form a multi-dimensional threshold matrix containing the corresponding thresholds of each core principal component.

[0012] Furthermore, the core principal components in the current optimized feature set are extracted and compared with the corresponding thresholds in the multi-dimensional threshold matrix. When all core principal components are less than the warning threshold, the state is determined to be fault-free. When there is a core principal component between the warning threshold and the severe fault threshold and no core principal component is greater than the severe fault threshold, the state is determined to be slightly chaotic. When there is any core principal component greater than the severe fault threshold, the state is determined to be severely chaotic.

[0013] A millimeter-wave fiber optic communication transmission fault detection system includes: The signal acquisition module installs signal couplers at the filter input and output ends of the millimeter-wave optical fiber communication link to generate the original reference signal and the signal to be detected, which are then modulated onto the optical carrier and output as a mixed signal after coherent superposition and balanced detection. The signal processing module performs detrending and noise filtering on the mixed signal to obtain chaotic feature signals. It calculates the time delay and embedding dimension through phase space reconstruction, constructs the phase space matrix, and determines the chaotic state of the system based on the maximum Lyapunov exponent. The feature extraction module calculates the correlation dimension and Kolmogorov entropy of the chaotic feature signals, constructs the feature matrix, and performs principal component analysis to obtain the optimized feature set. The fault determination module establishes a mapping relationship library between latency fluctuations and optimized feature sets, sets multi-level fault thresholds, and determines the fault level based on the comparison results between the current optimized feature set and the fault thresholds.

[0014] A storage medium storing one or more computer instructions for implementing the millimeter-wave optical fiber communication transmission fault detection method described above.

[0015] (III) Beneficial Effects This invention provides a method, system, and storage medium for detecting faults in millimeter-wave optical fiber communication transmission, which has the following beneficial effects: (1) By installing signal couplers at the input and output ends of the filter in the millimeter-wave optical fiber communication link, the original reference signal and the signal to be detected are generated. The two signals are modulated onto the optical carrier and output as a mixed signal through coherent superposition and balanced detection. This achieves high-precision signal acquisition and synchronous processing, ensures signal synchronization and amplitude consistency, effectively preserves sub-nanosecond phase jump characteristics, and improves signal purity through narrow linewidth laser and balanced detection technology. This provides a high-quality data foundation for subsequent chaotic analysis and significantly enhances the sensitivity and reliability of fault detection.

[0016] (2) The mixed signal is detrended and noise is filtered out by least squares method and wavelet threshold denoising method to obtain pure chaotic feature signal. The time delay and embedding dimension are calculated by combining phase space reconstruction technology to construct phase space matrix. The chaotic state of the system is determined based on the maximum Lyapunov exponent, effectively identifying the chaotic phenomenon caused by sub-nanosecond time delay fluctuation, improving the signal-to-noise ratio, accurately extracting chaotic features, optimizing reconstruction accuracy through dynamic parameters, enhancing the sensitivity and reliability of fault detection, and providing a high-quality data foundation for subsequent feature analysis.

[0017] (3) By calculating the correlation dimension and Kolmogorov entropy of chaotic feature signals, a feature matrix is ​​constructed and principal component analysis is performed to extract and optimize the feature set. The correlation dimension is used to quantify the system complexity, and the dynamic uncertainty is evaluated by Kolmogorov entropy. Combined with principal component analysis, feature dimensionality reduction is achieved and redundant information is eliminated. Principal components strongly correlated with faults are screened to improve the fault differentiation accuracy, provide highly discriminative feature basis for multi-level fault judgment, and improve the robustness and efficiency of detection.

[0018] (4) By establishing a mapping relationship library between time delay fluctuation and optimized feature set, and setting multi-level fault thresholds, the system can accurately classify and judge faults. Based on the measured data, a multi-dimensional threshold matrix is ​​constructed to support sub-nanosecond time delay fluctuation detection. The system can improve the accuracy by achieving three-level fault classification through core principal component comparison. Combined with the nearest neighbor matching algorithm, the system can trace the root cause of the fault, output real-time warnings or automatically switch to backup links, improve system reliability, and record complete fault data for analysis and optimization. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the steps of the millimeter-wave optical fiber communication transmission fault detection method of the present invention; Figure 2 This is a schematic diagram of the millimeter-wave optical fiber communication transmission fault detection method of the present invention; Figure 3 This is a schematic diagram of the millimeter-wave optical fiber communication transmission fault detection system of the present invention. Detailed Implementation

[0020] 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. 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.

[0021] Please see Figures 1-2 This invention provides a method for detecting faults in millimeter-wave optical fiber communication transmission, comprising the following steps: Step 1: Install signal couplers at the input and output ends of the filter in the millimeter-wave optical fiber communication link to generate the original reference signal and the signal to be detected, and modulate them onto the optical carrier respectively. After coherent superposition and balanced detection, the mixed signal is output. Step one includes the following: Step 101: Fix a first signal coupler at the filter input end of the millimeter-wave optical fiber communication transmission link, and fix a second signal coupler at the corresponding position at the filter output end. Set the coupling coefficient of the first and second signal couplers to 10%~20% to ensure that the amplitude deviation of the signals extracted by the two couplers does not exceed ±5% and does not affect the normal communication of the main link; select a high-precision clock synchronization module with an accuracy better than 1ps, and electrically connect it to the signal acquisition equipment of the two signal acquisition channels respectively to provide the two acquisition channels with the same source synchronization clock; configure a delay compensation unit with a fine-tuning accuracy of 0.1ns, connect it in series to the original reference signal acquisition channel, and adjust the transmission delay of the original reference signal through the delay compensation unit so that the time difference between the original reference signal and the signal to be detected reaching the subsequent signal processing module is less than 0.5ns, thus completing the synchronization calibration of the two channel signals; Step 102: Input the two-channel acquired signals after synchronous calibration into a programmable gain amplifier. Adjust the amplitude of the two-channel signals to the standard range of 0.5V~1V through the programmable gain amplifier to achieve signal amplitude normalization. Input the two-channel signals after amplitude normalization into a low-pass filter with a cutoff frequency of 100GHz. Filter out high-frequency noise interference with frequencies higher than 100GHz in the signal through the low-pass filter and output the original reference signal S1(t) and the signal to be detected S2(t) that are synchronously matched. Step 103: Select a narrow-linewidth laser with a linewidth of less than 10kHz, set its output wavelength to 1550nm, and start the laser to generate a continuous and stable optical carrier. Input the optical carrier into a 1×2 fiber beam splitter, which splits the optical carrier into two sub-optical carriers with equal power. The power deviation between the two sub-optical carriers is controlled within ±3%. Connect the two sub-optical carriers to the optical input terminals of the first electro-optic modulator and the second electro-optic modulator, respectively. At the same time, connect the original reference signal to the electrical input terminal of the first electro-optic modulator and the signal to be detected to the electrical input terminal of the second electro-optic modulator. Using an external modulation method, the original reference signal is loaded onto one sub-optical carrier through the first electro-optic modulator, and the signal to be detected is loaded onto the other sub-optical carrier through the second electro-optic modulator. The modulated optical signal and the modulated optical signal are output respectively. Step 104: Connect the modulated optical signal and the modulated optical signal to the two inputs of an optical fiber coupler with a splitting ratio of 1:1, respectively. Coherently superimpose the two modulated optical signals through the optical fiber coupler to output a coherent superimposed optical signal. Connect the coherent superimposed optical signal to the optical receiver of a balanced photodetector. Convert the coherent superimposed optical signal into an electrical signal through the balanced photodetector to obtain a beat frequency signal. The frequency of this beat frequency signal is equal to the frequency difference between the original reference signal and the signal to be detected, and it completely preserves the sub-nanosecond phase jump characteristics in the original signal. Input the beat frequency signal into a low-noise amplifier with a noise figure of less than 2dB. Amplify the beat frequency signal through the low-noise amplifier. Then input the amplified signal into an adaptive equalizer. The adaptive equalizer compensates for the signal distortion generated during the coherent mixing process and finally outputs the optimized mixed signal.

[0022] When using this method, refer to steps 101 to 104: By installing signal couplers at the input and output ends of the filter in the millimeter-wave optical fiber communication link, the original reference signal and the signal to be detected are generated. The two signals are then modulated onto an optical carrier and output as a mixed signal through coherent superposition and balanced detection. This achieves high-precision signal acquisition and synchronous processing, ensuring signal synchronization and amplitude consistency, effectively preserving sub-nanosecond phase transition characteristics, and improving signal purity through narrow-linewidth lasers and balanced detection technology. This provides a high-quality data foundation for subsequent chaotic analysis and significantly enhances the sensitivity and reliability of fault detection.

[0023] Step 2: Detrend and noise filtering are performed on the mixed signal to obtain the chaotic feature signal. The time delay and embedding dimension are calculated through phase space reconstruction to construct the phase space matrix. The chaotic state of the system is determined based on the maximum Lyapunov exponent. Step two includes the following: Step 201: Detrending the mixing signal is performed using the least squares method. By fitting the linear trend curve of the signal, this linear trend curve is removed from the mixing signal to eliminate the linear drift component in the signal. Noise is filtered out from the detrended signal using wavelet thresholding. The db4 wavelet basis is selected as the decomposition basis, and the number of decomposition levels is set to 3 to 5. The detrended signal is decomposed using db4 wavelet decomposition to obtain the wavelet coefficients of each level. The corresponding coefficient threshold is calculated based on the noise of each level of signal. The coefficient threshold is equal to the product of the noise standard deviation and the square root of twice the natural logarithm of the signal length. Thresholding is performed on the wavelet coefficients of each level, setting wavelet coefficients with absolute values ​​less than or equal to the coefficient threshold to zero and retaining wavelet coefficients with absolute values ​​greater than the coefficient threshold. The processed wavelet coefficients of each level are then subjected to inverse db4 wavelet transform to complete signal reconstruction, resulting in a clean chaotic feature signal. Step 202: Calculate the time delay and embedding dimension required for phase space reconstruction. The time delay is determined by calculating the autocorrelation function of the chaotic feature signal. The time value corresponding to the first drop of the autocorrelation function to 1 / e of its maximum value is taken as the time delay. The embedding dimension is determined by the false neighbor method. The embedding dimension is gradually increased, and the proportion of false neighbors in each dimension is calculated. When the proportion of false neighbors drops below 5% for the first time, the corresponding dimension is the final embedding dimension. Based on the determined time delay and embedding dimension, the phase space matrix is ​​constructed using the delayed coordinate method. This phase space matrix consists of vectors of multiple embedding dimensions. Each embedding dimension vector is composed of sampled values ​​from different times in the chaotic feature signal arranged in order of time delay, and the starting sampling times of adjacent vectors increase sequentially. Calculate the product of the embedding dimension minus one and the time delay. The total number of points in the reconstructed phase space is equal to the total number of sampled points of the chaotic feature signal minus this product. Step 203: Randomly select an initial reference point from the phase space matrix, and find a nearest neighbor point that does not coincide with the initial reference point within a small distance neighborhood of the initial reference point. The small distance threshold ranges from 0.01 to 0.05 times the peak value of the chaotic feature signal. Calculate the Euclidean distance between the initial reference point and the nearest neighbor point, and record it as the initial distance. Track the evolution trajectory of the initial reference point and the nearest neighbor point over time, and calculate the Euclidean distance between the two corresponding evolution points at different evolution times. Take the natural logarithm of the distance values ​​at all evolution times, and perform linear fitting with the number of evolution steps as the abscissa and the corresponding natural logarithm value as the ordinate. The slope of the fitted line is the maximum Lyapunov exponent. When the maximum Lyapunov exponent is greater than 0, it is determined that the system corresponding to the chaotic feature signal has a chaotic phenomenon, and the larger the value of the exponent, the higher the chaotic intensity of the system.

[0024] When using this method, refer to steps 201 to 203: The mixed signal is detrended and noise filtered by least squares and wavelet threshold denoising to obtain a clean chaotic feature signal. The time delay and embedding dimension are calculated by combining phase space reconstruction technology to construct the phase space matrix. The chaotic state of the system is determined based on the maximum Lyapunov exponent, which effectively identifies chaotic phenomena caused by sub-nanosecond time delay fluctuations, improves the signal-to-noise ratio, and accurately extracts chaotic features. The reconstruction accuracy is optimized by dynamic parameters to enhance the sensitivity and reliability of fault detection and provide a high-quality data foundation for subsequent feature analysis.

[0025] Step 3: Calculate the correlation dimension and Kolmogorov entropy of the chaotic feature signals, construct the feature matrix, and perform principal component analysis to obtain the optimized feature set; Step three includes the following: Step 301: Set the scale range to 0.01 to 0.5 times the peak amplitude of the chaotic feature signal, and uniformly select 20 to 50 different scale values ​​within this scale range; for each scale value, construct the corresponding correlation integral: count the number of point pairs in all state point pairs in the phase space whose distance between two points is ≤ the current scale value, and take the ratio of this number of point pairs to the total number of all possible state point pairs in the phase space. This ratio is the correlation integral at the current scale; after completing the construction of the correlation integrals at all scales, perform a linear fit on the logarithmic relationship between the correlation integral and the scale. The slope of the fitted line is the correlation dimension; based on the time delay, select a delay time range of 1 to 5 times the time delay, calculate the mutual information value of the signal at each delay time, and perform a linear fit on the decay rate of the mutual information value with the delay. The absolute value of the slope of the fitted line is the Kolmogorov entropy; Step 302: Organize all chaotic feature parameters into a feature matrix. The chaotic feature parameters include the maximum Lyapunov exponent, correlation dimension, and Kolmogorov entropy. Each row corresponds to one set of samples, and each column corresponds to one feature parameter. The feature matrix is ​​standardized using Z-score, that is, each feature value is subtracted from its mean and then divided by its standard deviation to eliminate the dimensional differences between different features. Then, the covariance matrix of the standardized feature matrix is ​​calculated, and the Jacobi method is used to solve for the eigenvalues ​​and corresponding eigenvectors of the covariance matrix. The variance contribution rate of each feature is calculated according to the magnitude of the eigenvalues, that is, the ratio of a single eigenvalue to the sum of all eigenvalues. The variance contribution rates are accumulated in descending order of eigenvalues, and the eigenvectors corresponding to the top k features with a cumulative variance contribution rate of 85% or higher are selected as principal components. The standardized feature matrix is ​​multiplied by the matrix composed of the selected k principal components to obtain the optimized feature set after dimensionality reduction. Step 303: Construct a validation dataset. This dataset must contain complete feature data corresponding to three types of signals: feature data under normal operation of the millimeter-wave fiber optic communication link; feature data of slight chaotic faults caused by filter delay fluctuations (delay fluctuation range 0.1 to 0.4 ns); and feature data of severe chaotic faults caused by filter delay fluctuations (delay fluctuation range 0.6 to 1.0 ns). The sample size for each type of data should be no less than 100 sets, and the samples should cover the collected data at different operating temperatures, from -20℃ to 60℃. Calculate the Pearson correlation coefficient between each principal component in the optimized feature set and the fault severity, quantizing normal, slight chaotic faults, and severe chaotic faults as 0, 1, and 2, respectively. Principal components with absolute correlation coefficients less than the correlation threshold (default threshold is 0.7) are removed, while effective features strongly correlated with fault severity are retained. Cross-validation is used to verify the fault discrimination capability of the optimized feature set. The validation dataset is randomly divided into training and test sets in a 7:3 ratio. A support vector machine (SVM) fault discrimination model is trained using the training set with radial basis functions as the kernel function. The model's discrimination accuracy is tested using the test set, requiring a test accuracy of no less than 90%. If the requirement is not met, the process returns to step 302 to readjust the principal component selection criteria. The adjustment direction is to increase the cumulative variance contribution rate threshold, such as adjusting it to 90%, or to increase the number of principal components, until the model's test accuracy meets the requirements.

[0026] When using this method, refer to the content of steps 301 to 303: By calculating the correlation dimension and Kolmogorov entropy of chaotic feature signals, a feature matrix is ​​constructed and principal component analysis is performed to extract and optimize the feature set. The correlation dimension is used to quantify the system complexity, and the Kolmogorov entropy is used to evaluate dynamic uncertainty. Combined with principal component analysis, feature dimensionality reduction is achieved to eliminate redundant information. Principal components strongly correlated with faults are screened to improve the fault differentiation accuracy, provide highly discriminative feature basis for multi-level fault determination, and improve the robustness and efficiency of detection.

[0027] Step 4: Establish a mapping relationship library between latency fluctuations and optimized feature sets, set multi-level fault thresholds, and determine the fault level based on the comparison results between the current optimized feature set and the fault thresholds.

[0028] Step four includes the following: Step 401: Connect the filter consistent with the actual application to the fault simulation platform and set the simulation parameters: the filter delay fluctuation range is adjusted from 0.1ns to 1.0ns, divided into multiple gradients at 0.05ns intervals; the operating temperature range is adjusted from -20℃ to 60℃, divided into multiple temperature points at 10℃ intervals; each delay fluctuation gradient and each temperature point is combined into a test condition, and 30 sets of pure chaotic feature signals are collected under each condition. After processing in Steps 2 and 3, the corresponding optimized feature set is obtained; the filter delay fluctuation, optimized feature set, and operating temperature data of all test conditions are organized into a mapping relationship library, and each data in the library... Each value includes a unique latency fluctuation, the corresponding complete optimized feature set data, and the test temperature. Multi-level fault thresholds are calibrated: the maximum value of the core principal components of the optimized feature set corresponding to fault-free operating conditions in the statistical mapping relation library is calculated, and the latency fluctuation of fault-free operating conditions is <0.1ns, so this value is increased by 5% as the warning threshold; the minimum value of the core principal components of the optimized feature set corresponding to severe fault operating conditions is calculated, and the latency fluctuation of severe fault operating conditions is ≥0.6ns, so this value is decreased by 5% as the severe fault threshold; the interval between the warning threshold and the severe fault threshold is the minor fault threshold interval, ultimately forming a multi-dimensional threshold matrix containing the warning threshold and severe fault threshold of each core principal component. Step 402: Obtain the optimized feature set of the current detection link after processing in Steps 2 and 3, and extract the core principal components; compare each core principal component with the corresponding warning threshold and severe fault threshold in the threshold matrix; if the values ​​of all core principal components are less than the corresponding warning threshold, the current link is determined to be in a fault-free state; if the values ​​of some core principal components are between the corresponding warning threshold and the severe fault threshold, and no core principal component value is greater than the severe fault threshold, the current link is determined to be in a slightly chaotic fault state; if the value of any core principal component is greater than the corresponding severe fault threshold, the current link is determined to be in a severely chaotic fault state. Step 403: Input the currently detected optimized feature set into the mapping relationship library established in Step 401. Use the nearest neighbor matching algorithm to calculate the Euclidean distance between the current optimized feature set and each data optimized feature set in the library. Select the data with the smallest distance as the matching result. Based on the time delay fluctuation and fault condition information corresponding to the matching result, trace and confirm that the root cause of the fault is the filter time delay fluctuation, and estimate the actual time delay fluctuation of the current filter. Take the time delay fluctuation from the matching result. In the fault-free state, output a green normal indicator signal and simultaneously upload the link normal status information. In the slightly chaotic fault state, output a yellow warning indicator signal and simultaneously upload the warning information of slight filter time delay fluctuation, estimated fluctuation, and suggestion to strengthen real-time monitoring. In the severely chaotic fault state, output a red alarm indicator signal, immediately trigger the preset link switching command, switch to the backup communication link, and simultaneously upload the alarm information of severe filter time delay fluctuation, estimated fluctuation, and triggered link switching. Automatically record the fault occurrence time, the specific values ​​of each core principal component in the optimized feature set, the estimated time delay fluctuation, the current link operating temperature, and other data, store them in the local database and back them up synchronously.

[0029] When using this method, please refer to the content of steps 401 to 403: By establishing a mapping relationship library between latency fluctuations and optimized feature sets, and setting multi-level fault thresholds, accurate fault classification and judgment can be achieved. A multi-dimensional threshold matrix is ​​constructed based on measured data to support sub-nanosecond latency fluctuation detection. The accuracy is improved by achieving three-level fault classification through core principal component comparison. Combined with the nearest neighbor matching algorithm, the root cause of the fault is traced, and real-time early warning or automatic switching to backup links is output to improve system reliability. At the same time, complete fault data is recorded for analysis and optimization.

[0030] Please see Figure 3 The present invention also provides a millimeter-wave optical fiber communication transmission fault detection system, comprising: a signal acquisition module, a signal processing module, a feature extraction module, and a fault determination module, wherein: The signal acquisition module installs signal couplers at the filter input and output ends of the millimeter-wave optical fiber communication link to generate the original reference signal and the signal to be detected, which are then modulated onto the optical carrier and output as a mixed signal after coherent superposition and balanced detection. The signal processing module performs detrending and noise filtering on the mixed signal to obtain chaotic feature signals. It calculates the time delay and embedding dimension through phase space reconstruction, constructs the phase space matrix, and determines the chaotic state of the system based on the maximum Lyapunov exponent. The feature extraction module calculates the correlation dimension and Kolmogorov entropy of the chaotic feature signals, constructs the feature matrix, and performs principal component analysis to obtain the optimized feature set. The fault determination module establishes a mapping relationship library between latency fluctuations and optimized feature sets, sets multi-level fault thresholds, and determines the fault level based on the comparison results between the current optimized feature set and the fault thresholds.

[0031] The present invention also provides a storage medium storing one or more computer instructions, the one or more computer instructions being used to implement the millimeter-wave optical fiber communication transmission fault detection method provided by the present invention.

[0032] In the application, the various formulas mentioned are all calculated by removing dimensions and taking their numerical values. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The coefficients in the formulas are set by those skilled in the art according to the actual situation.

[0033] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, and combinations thereof. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0034] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0035] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for detecting faults in millimeter-wave optical fiber communication transmission, characterized in that, include: Signal couplers are installed at the input and output ends of the filter in the millimeter-wave optical fiber communication link to generate the original reference signal and the signal to be detected, which are then modulated onto the optical carrier. After coherent superposition and balanced detection, the mixed signal is output. The mixed signal is detrended and noise filtered to obtain chaotic feature signals. The time delay and embedding dimension are calculated through phase space reconstruction to construct the phase space matrix. The chaotic state of the system is determined based on the maximum Lyapunov exponent. Calculate the correlation dimension and Kolmogorov entropy of the chaotic feature signals, construct the feature matrix, and perform principal component analysis to obtain the optimized feature set; Establish a mapping relationship library between latency fluctuations and optimized feature sets, set multi-level fault thresholds, and determine the fault level based on the comparison results between the current optimized feature set and the fault thresholds.

2. The method for detecting millimeter-wave optical fiber communication transmission faults according to claim 1, characterized in that, The signal couplers installed at the input and output of the filter are, in sequence, a first signal coupler and a second signal coupler. The clock synchronization module provides a synchronous clock for the two signal acquisition channels, and the delay compensation unit completes the synchronization calibration of the two acquired signals. The amplitude of the two acquired signals after synchronization calibration is normalized, and high-frequency noise is filtered out by a low-pass filter to obtain the original reference signal and the signal to be detected.

3. The method for detecting millimeter-wave optical fiber communication transmission faults according to claim 2, characterized in that, A continuous optical carrier is generated by a narrow-linewidth laser. The optical carrier is input into a 1×2 fiber beam splitter and split into two sub-optical carriers. The original reference signal and the signal to be detected are loaded onto the two sub-optical carriers by external modulation, resulting in two modulated optical signals. The two modulated optical signals are coherently superimposed by a fiber coupler with a splitting ratio of 1:1, converted into electrical signals, and then processed by a low-noise amplifier and an adaptive equalizer to output a mixed signal.

4. The method for detecting millimeter-wave optical fiber communication transmission faults according to claim 1, characterized in that, The least squares method is used to detrend the mixed signal to eliminate the linear drift component. Wavelet thresholding is then used to filter out noise from the detrended mixed signal, resulting in a clean chaotic feature signal. The time delay and embedding dimension required for phase space reconstruction are calculated. The time delay is determined by calculating the autocorrelation function of the chaotic feature signal, and the time value corresponding to the first drop of the autocorrelation function to 1 / e of its maximum value is taken as the time delay. The embedding dimension is determined by the spurious neighbor method, gradually increasing the embedding dimension while calculating the proportion of spurious neighbors in each dimension. When the proportion of spurious neighbors first drops below 5%, the corresponding dimension is the final embedding dimension. Based on the determined time delay and embedding dimension, the phase space matrix is ​​constructed using the delay coordinate method.

5. The method for detecting millimeter-wave optical fiber communication transmission faults according to claim 4, characterized in that, By randomly selecting an initial reference point from the phase space matrix, a nearest neighbor point that does not coincide with the initial reference point is found within a small distance neighborhood of the initial reference point. The Euclidean distance between the initial reference point and the nearest neighbor point is calculated and denoted as the initial distance. The evolution trajectories of the initial reference point and the nearest neighbor point over time are tracked, and the Euclidean distance between the two corresponding evolution points at different evolution times is calculated sequentially. The natural logarithm of the distance values ​​at all evolution times is taken, and a linear fit is performed with the number of evolution steps as the x-axis and the corresponding natural logarithm value as the y-axis. The slope of the fitted line is the maximum Lyapunov exponent. When the maximum Lyapunov exponent is greater than 0, it is determined that the system corresponding to the chaotic feature signal has a chaotic phenomenon.

6. The method for detecting millimeter-wave optical fiber communication transmission faults according to claim 1, characterized in that, Within a defined scale range, a correlation integral is constructed, and the correlation dimension is obtained through linear fitting. The mutual information value of the signal at each delay time is calculated based on the time delay, and the Kolmogorov entropy is obtained by fitting the decay rate of the mutual information value. The maximum Lyapunov exponent, correlation dimension, and Kolmogorov entropy are organized into a feature matrix, and after standardization, principal component analysis is performed. Principal components with a cumulative variance contribution rate of 85% or higher are selected to form an optimized feature set. Principal components whose absolute values ​​of correlation coefficients with fault severity are less than the correlation threshold are removed. The fault discrimination capability of the optimized feature set is verified by cross-validation until the preset accuracy is met.

7. The method for detecting millimeter-wave optical fiber communication transmission faults according to claim 1, characterized in that, Obtain optimized feature sets under different combinations of time delay fluctuation gradients and operating temperatures, and organize them into a mapping relationship library between time delay fluctuations and optimized feature sets; calibrate the warning threshold, minor fault threshold range, and severe fault threshold, and form a multi-dimensional threshold matrix containing the corresponding thresholds of each core principal component.

8. The method for detecting millimeter-wave optical fiber communication transmission faults according to claim 7, characterized in that, Extract the core principal components from the current optimized feature set and compare them with the corresponding thresholds in the multi-dimensional threshold matrix; when all core principal components are less than the warning threshold, the state is determined to be fault-free. When there is a core principal component between the warning threshold and the severe fault threshold and no core principal component is greater than the severe fault threshold, it is judged as a slight chaotic fault state. If any core principal component is greater than the severe fault threshold, the state is determined to be a severe chaotic fault state.

9. A millimeter-wave optical fiber communication transmission fault detection system, characterized in that, include: The signal acquisition module installs signal couplers at the filter input and output ends of the millimeter-wave optical fiber communication link to generate the original reference signal and the signal to be detected, which are then modulated onto the optical carrier and output as a mixed signal after coherent superposition and balanced detection. The signal processing module performs detrending and noise filtering on the mixed signal to obtain chaotic feature signals. It calculates the time delay and embedding dimension through phase space reconstruction, constructs the phase space matrix, and determines the chaotic state of the system based on the maximum Lyapunov exponent. The feature extraction module calculates the correlation dimension and Kolmogorov entropy of the chaotic feature signals, constructs the feature matrix, and performs principal component analysis to obtain the optimized feature set. The fault determination module establishes a mapping relationship library between latency fluctuations and optimized feature sets, sets multi-level fault thresholds, and determines the fault level based on the comparison results between the current optimized feature set and the fault thresholds.

10. A storage medium, characterized in that, The storage medium stores one or more computer instructions for implementing the method as described in any one of claims 1 to 8.

Images