Optical fiber link anomaly analysis method and electronic device

Abstract

This application relates to the field of optical fiber communication technology, and discloses a method and electronic device for optical fiber link anomaly analysis. The method includes: acquiring a modulated optical signal from the transmitting end of the optical fiber link and an actual sampled signal from the receiving end of the optical fiber link; generating an actual observation signal based on the actual sampled signal; constructing a differentiable physical model; inputting the modulated optical signal into the differentiable physical model to generate a predicted observation signal; iteratively optimizing the distribution of fiber dispersion parameters and longitudinal nonlinear parameters in the differentiable physical model based on the predicted observation signal and the actual observation signal until the differentiable physical model converges; and performing anomaly analysis on the optical fiber link based on the converged longitudinal nonlinear parameter distribution. This application can improve the accuracy of optical fiber link anomaly analysis.

Description

Technical Field

[0001] This application relates to the field of optical fiber communication technology, and in particular to an optical fiber link anomaly analysis method and electronic device. Background Technology

[0002] With the continuous development of high-speed, high-capacity, and long-distance optical fiber communication systems, the importance of optical fiber link transmission status monitoring and fault diagnosis is increasing. In actual links, local abnormal additional losses may occur due to connector or fusion splice deterioration, bending, or compression, leading to a decline in transmission performance. Therefore, how to monitor key state quantities along the length of the optical fiber link without interrupting services, and accurately locate and estimate abnormal losses, has become an important research topic in the field of optical fiber communication link operation and maintenance and intelligent diagnosis.

[0003] In related technologies, perturbation theory-based longitudinal power estimation methods for fiber optic links are used for anomaly analysis. These methods utilize perturbation information in the received signal to invert the longitudinal power distribution of the link. However, these methods are highly dependent on prior link parameters, especially dispersion parameters, which are particularly sensitive. Generally, it is assumed that the dispersion parameters are known or accurately preset. When the actual dispersion parameters deviate from the preset values, a mismatch occurs between the established model and the actual transmission process, leading to a decrease in the accuracy of longitudinal power distribution estimation and affecting the location and amplitude estimation results of abnormal additional losses. Summary of the Invention

[0004] The purpose of this application is to provide a method and electronic device for analyzing fiber optic link anomalies, and to improve the accuracy of fiber optic link anomaly analysis.

[0005] To address the aforementioned technical problems, embodiments of this application provide a method for analyzing fiber optic link anomalies, comprising: Acquire the modulated optical signal from the transmitting end of the optical fiber link, and the actual sampled signal from the receiving end of the optical fiber link; The actual observation signal is generated based on the actual sampled signal; A differentiable physical model is constructed, wherein the parameters to be optimized in the differentiable physical model include the fiber dispersion parameters and longitudinal nonlinear parameter distributions of the fiber link, and the output of the differentiable physical model is the predicted observation signal corresponding to the actual observation signal. The modulated optical signal is input into the differentiable physical model to generate the predicted observation signal; Based on the predicted observation signal and the actual observation signal, the distribution of the fiber dispersion parameter and the longitudinal nonlinear parameter in the differentiable physical model is iteratively optimized until the differentiable physical model converges; Anomaly analysis is performed on the optical fiber link based on the converged longitudinal nonlinear parameter distribution.

[0006] In an optional embodiment, the fiber dispersion parameter is the second-order group velocity dispersion coefficient of the fiber link; the longitudinal nonlinear parameter distribution is a nonlinear weight coefficient vector that is equally spaced in the transmission direction of the fiber link, wherein each component of the nonlinear weight coefficient vector corresponds one-to-one with a longitudinal segment of the fiber link.

[0007] In an optional embodiment, the step of iteratively optimizing the distributions of the fiber dispersion parameter and the longitudinal nonlinear parameter in the differentiable physical model based on the predicted observation signal and the actual observation signal until the differentiable physical model converges includes: Construct an objective function for the error between the predicted observed signal and the actual observed signal; The gradient of the error objective function with respect to the fiber dispersion parameter and the longitudinal nonlinear parameter distribution is calculated based on the automatic differentiation mechanism. The fiber dispersion parameter and the longitudinal nonlinear parameter distribution are iteratively updated until the error objective function converges.

[0008] In an optional embodiment, the formula for determining the error objective function includes: ; in, A [ L [Length is] L The actual observed signal of the aforementioned fiber optic link, A pred [ L ] represents the predicted observation signal output by the differentiable physical model, and γ′ represents the distribution of the longitudinal nonlinear parameters. R (γ′) is a regularization term for the longitudinal nonlinear parameter distribution.

[0009] In an optional embodiment, the iterative update of the fiber dispersion parameters and the longitudinal nonlinear parameter distributions includes: The distributions of the fiber dispersion parameters and the longitudinal nonlinear parameters are jointly iteratively updated based on the Adam optimization algorithm.

[0010] In an optional embodiment, constructing a differentiable physical model includes: A linear dispersion calculation branch is constructed, which is used to calculate the linear ideal optical field of the modulated optical signal at the receiving end of the optical fiber link based on the frequency domain dispersion transmission operator, and to calculate the piecewise linear optical field of the modulated optical signal at each longitudinal segment position of the optical fiber link. A nonlinear perturbation calculation branch is constructed, which is used to calculate the nonlinear perturbation kernel coefficients corresponding to each segment based on the piecewise linear optical field, so as to construct a nonlinear perturbation matrix according to the nonlinear perturbation kernel coefficients; Based on the superposition relationship between the linear ideal optical field and the nonlinear perturbation matrix, a mapping function is established between the predicted observation signal and the distribution of the fiber dispersion parameter and the longitudinal nonlinear parameter.

[0011] In an optional embodiment, the formula for calculating the linear ideal light field includes: ; The calculation formula for the piecewise linear optical field includes: ; The formula for calculating the nonlinear perturbation kernel coefficients includes: ; The formula for calculating the nonlinear perturbation matrix includes: ; The formula for determining the mapping function includes: ; Where, A0[ [ ] represents the linear ideal light field. L The total transmission length of the optical fiber link is given, A[0] is the modulated optical signal, β2 is the optical fiber dispersion parameter, ω is the angular frequency, and n0[ ] represents the piecewise linear optical field. Let z be the distance from the transmitter of the optical fiber link to the k-th longitudinal segment position. k The transmission length, FFT( ) is the Fast Fourier Transform, IFFT( ) represents the inverse fast Fourier transform, and Δz represents the unit step size of the longitudinal segmentation of the optical fiber link. The nonlinear perturbation kernel coefficients are... For from the first k The vertical segment position z k The frequency domain dispersion transmission operator to the receiving end of the optical fiber link, N ( ) is a nonlinear mapping operator. Let m be the nonlinear perturbation matrix. z γ is the total number of segments in the optical fiber link; γ′ is the longitudinal nonlinear parameter distribution, A pred [ ] represents the predicted observation signal.

[0012] In an optional embodiment, the anomaly analysis of the optical fiber link based on the converged longitudinal nonlinear parameter distribution includes: The converged longitudinal nonlinear parameter distribution is mapped to the optical power value at each longitudinal position; The longitudinal optical power curve of the optical fiber link is generated based on the optical power value; Identify the locations of local abrupt changes in the longitudinal optical power curve to determine the locations of abnormal losses, and determine the loss amplitude of the optical fiber link based on the power difference before and after the locations of the local abrupt changes.

[0013] In an optional embodiment, generating the actual observation signal based on the actual sampled signal includes: The actual sampled signal is sequentially resampled, dispersion pre-compensated, polarization demultiplexed, and carrier phase recovered to obtain the baseband processed signal; The baseband processed signal is reloaded with the compensated linear dispersion to generate the actual observed signal.

[0014] Embodiments of this application also provide an electronic device, including: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method as described above.

[0015] The fiber optic link anomaly analysis method provided in this application constructs a differentiable physical model based on the distribution of fiber dispersion parameters and longitudinal nonlinear parameters. After actually acquiring the modulated optical signal from the transmitting end of the fiber optic link and the actual sampled signal from the receiving end, the modulated optical signal is input into the differentiable physical model. The differentiable physical model predicts the signal at the receiving end. The predicted observed signal is compared with the actual observed signal obtained by processing the actual sampled signal, thereby iteratively optimizing the differentiable physical model. Since the model iterative optimization process is essentially an optimization adjustment of the model parameters, namely the fiber dispersion parameters and the longitudinal nonlinear parameter distribution, the final convergence of the differentiable physical model indicates that the determined fiber dispersion parameters and longitudinal nonlinear parameter distribution better match the actual fiber optic link situation. This allows for more accurate fiber optic link anomaly analysis. Therefore, by transforming the fiber dispersion parameters from fixed prior values ​​into optimizable variables, this application can, to some extent, solve the dispersion mismatch problem of traditional perturbation theory methods. It can achieve accurate inversion of the longitudinal optical power distribution of the link without precise link prior parameters, significantly improving the intelligence level of fiber optic link operation and maintenance and the accuracy of fault diagnosis. Attached Figure Description

[0016] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0017] Figure 1 This is a flowchart illustrating the fiber optic link anomaly analysis method provided in one embodiment of this application; Figure 2 This is a flowchart illustrating the fiber optic link anomaly analysis method provided in another embodiment of this application; Figure 3 This is a schematic diagram of the convergence of fiber dispersion parameters during iterative optimization of a differentiable physical model in one embodiment of this application. Figure 4 This is a schematic diagram illustrating the iterative optimization of a differentiable physical model based on the signals from the transmitting and receiving ends of an optical fiber link in one embodiment of this application. Figure 5 This is a flowchart illustrating the fiber optic link anomaly analysis method provided in another embodiment of this application; Figure 6 This is a schematic diagram of the longitudinal optical power curve when there is local abnormal additional loss in the optical fiber link in one embodiment of this application; Figure 7 This is a schematic diagram of the structure of an electronic device provided in one embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the various embodiments of this application will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details have been provided in the various embodiments of this application to help readers better understand this application. However, the technical solutions claimed in this application can be implemented even without these technical details and various changes and modifications based on the following embodiments. The division of the various embodiments below is for the convenience of description and should not constitute any limitation on the specific implementation of this application. The various embodiments can be combined with and referenced by each other without contradiction.

[0019] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0020] To address the aforementioned technical problems, one embodiment of this application proposes a method for analyzing fiber optic link anomalies, referring to... Figure 1 As shown, the method includes: Step 100: Obtain the modulated optical signal from the transmitting end of the optical fiber link and the actual sampled signal from the receiving end of the optical fiber link.

[0021] The modulated optical signal is the baseband modulated signal to be transmitted generated at the transmitting end of the optical fiber link. In this embodiment, it can be a 150G Baud 16QAM modulated optical signal, or other higher-order modulation formats such as QPSK or 64QAM can be used according to actual system requirements. This signal also serves as the input reference signal for the differentiable physical model and is completely consistent with the service signal sent into the optical fiber link by the transmitting end. The actual sampled signal is the baseband electrical signal acquired by the coherent receiver at the receiving end after coherent detection and analog-to-digital conversion, and transmitted through the target optical fiber link.

[0022] In one optional embodiment, for a multi-span fiber optic link, the transmitter reference signal and receiver sample signal of each span can be obtained and the link anomaly analysis can be performed separately. Alternatively, the multi-span link can be treated as a whole for full-link analysis. No absolute limitation is imposed here.

[0023] Step 200: Generate the actual observation signal based on the actual sampled signal.

[0024] This step allows for preprocessing of the actual sampled signals during subsequent analysis to remove signals irrelevant to link parameter estimation from the received signals, resulting in clean observation samples that fit the differentiable physical model.

[0025] Specifically, in an optional embodiment, generating the actual observation signal based on the actual sampled signal includes: The actual sampled signal is sequentially resampled, dispersion pre-compensated, polarization demultiplexed, and carrier phase recovered to obtain the baseband processed signal; The baseband processed signal is reloaded with the compensated linear dispersion to generate the actual observed signal.

[0026] Resampling involves resampling the actual sampled signal acquired at the receiving end to a sampling rate consistent with the modulated optical signal at the transmitting end, thus aligning the sampling rates of the transmitted and received signals. Dispersion pre-compensation involves performing preliminary dispersion compensation on the received signal through digital signal processing. The compensation amount is the nominal dispersion value of the fiber optic link, providing a basis for subsequent polarization demultiplexing and carrier phase recovery. Polarization demultiplexing involves using a constant mode algorithm to separate the two orthogonal polarization states in the received signal, solving the signal aliasing problem caused by polarization mode dispersion in fiber transmission. Carrier phase recovery compensates for carrier phase noise caused by laser linewidth, recovering a clean baseband modulated signal. Dispersion reloading involves reloading all the previously pre-compensated linear dispersion onto the pre-processed baseband signal, recovering a signal consistent with the real optical field at the receiving end of the fiber optic link. This signal serves as the actual observation signal for parameter estimation, ensuring the physical consistency between the input of the differentiable physical model and the observation signal.

[0027] Step 300: Construct a differentiable physical model, wherein the parameters to be optimized in the differentiable physical model include the fiber dispersion parameters and longitudinal nonlinear parameter distributions of the fiber link, and the output of the differentiable physical model is the predicted observation signal corresponding to the actual observation signal.

[0028] Specifically, the differentiable physical model can be constructed based on the first-order perturbation theory of optical fiber transmission. Unlike traditional black-box neural network models, all operators and computational processes in this model strictly follow the physical laws of optical fiber transmission, making it a physics-driven white-box model. The input to the differentiable physical model is the modulated optical signal from the transmitter, and the output is a predicted observation signal corresponding to the actual observed signal. The differentiable physical model outputs the predicted observation signal based on the mapping function between the optical fiber dispersion parameters, the longitudinal nonlinear parameter distribution, and the observed signal. In this application, the optical fiber dispersion parameters are not considered as fixed, known prior parameters, but rather, together with the longitudinal nonlinear parameter distribution, they are used as target variables to be optimized and incorporated into a unified model framework. This solves the model mismatch problem caused by the deviation between the preset dispersion value and the actual value in traditional schemes.

[0029] In an optional embodiment, the differentiable physical model can be built on a deep learning framework that supports automatic differentiation, such as PyTorch or TensorFlow. All operators are implemented using the framework's native differentiable operators, eliminating the need to manually derive gradient formulas and significantly reducing the difficulty of model implementation.

[0030] Step 400: Input the modulated optical signal into the differentiable physical model to generate the predicted observation signal.

[0031] It can be understood that by inputting the modulated optical signal from the transmitting end into the constructed differentiable physical model, and based on the values ​​of the fiber dispersion parameters and longitudinal nonlinear parameter distributions at the current iteration step, the model can complete forward calculations to output the predicted observation signal corresponding to the current parameters.

[0032] Step 500: Based on the predicted observation signal and the actual observation signal, iteratively optimize the distribution of the fiber dispersion parameter and the longitudinal nonlinear parameter in the differentiable physical model until the differentiable physical model converges.

[0033] It can be understood that this step is the reverse solution process of the model. By using the fitting error between the predicted and actual observed signals, the two parameters to be optimized are iteratively optimized in reverse, so that the predicted signal output by the model fits the real signal as closely as possible, thereby obtaining the parameter estimation results that match the real link state, so as to complete the convergence of the differentiable physical model.

[0034] Step 600: Perform anomaly analysis on the optical fiber link based on the converged longitudinal nonlinear parameter distribution.

[0035] Based on the model after iterative convergence, the determined longitudinal nonlinear parameters can reflect the real link state to a certain extent. Therefore, the longitudinal transmission state of the optical fiber link can be recovered by inversion based on the distribution of the converged longitudinal nonlinear parameters, thereby completing the anomaly analysis of the optical fiber link.

[0036] Therefore, the fiber optic link anomaly analysis method provided in this application constructs a differentiable physical model based on the distribution of fiber dispersion parameters and longitudinal nonlinear parameters. After actually acquiring the modulated optical signal at the transmitting end of the fiber optic link and the actual sampled signal at the receiving end, the modulated optical signal is input into the differentiable physical model. The differentiable physical model predicts the signal at the receiving end. Based on the comparison between the predicted observed signal and the actual observed signal obtained by processing the actual sampled signal, the differentiable physical model is iteratively optimized. Since the model iterative optimization process is actually an optimization and adjustment of the model parameters, namely the distribution of fiber dispersion parameters and longitudinal nonlinear parameters, the fiber dispersion parameters and longitudinal nonlinear parameter distribution determined at the final convergence of the differentiable physical model can better match the actual fiber optic link situation. This allows for more accurate fiber optic link anomaly analysis. Thus, by transforming the fiber dispersion parameters from fixed prior values ​​into optimizable variables, this application can solve the dispersion mismatch problem of traditional perturbation theory methods to a certain extent. It can achieve accurate inversion of the longitudinal optical power distribution of the link without precise link prior parameters, greatly improving the intelligence level of fiber optic link operation and maintenance and the accuracy of fault diagnosis.

[0037] In an optional embodiment of this application, the fiber dispersion parameter is the second-order group velocity dispersion coefficient of the fiber link; the longitudinal nonlinear parameter distribution is a nonlinear weight coefficient vector that is equally spaced in the transmission direction of the fiber link, wherein each component of the nonlinear weight coefficient vector corresponds one-to-one with a longitudinal segment of the fiber link.

[0038] Specifically, the fiber dispersion parameter is the second-order group velocity dispersion coefficient β2 of the fiber, which is the core physical parameter describing the chromatic dispersion (CD) of the fiber. It determines the extent to which the light pulse is broadened in the fiber and is the core source of linear transmission damage in the fiber. The unit is ps² / km. In traditional schemes, this parameter is a fixed preset value, but in this embodiment, it is a continuous variable that can be iteratively optimized.

[0039] The longitudinal nonlinear parameter distribution is specifically a nonlinear weight coefficient vector γ′ set at equal intervals along the transmission direction of the optical fiber link. In the specific implementation process, the optical fiber link with a total length of L can be divided into mz longitudinal segments, and the length of each segment is Δz=L / mz. Each segment corresponds to an independent component γk′ (k=1,2,...,mz) in the vector γ′. This component is used to characterize the intensity of the nonlinear effect at the corresponding segment position and corresponds one-to-one with the optical fiber segment position.

[0040] In an optional embodiment, the nonlinear weight coefficient vector can also be set using an unequal interval segmentation method according to the actual link structure such as the segment division of the fiber optic link and the location of amplifiers, without any absolute limitation.

[0041] Reference Figure 2 As shown, in an optional embodiment of this application, the step of iteratively optimizing the distributions of the fiber dispersion parameter and the longitudinal nonlinear parameter in the differentiable physical model based on the predicted observation signal and the actual observation signal until the differentiable physical model converges includes: Step 501: Construct the objective function of the error between the predicted observed signal and the actual observed signal; Step 502: Calculate the gradient of the error objective function with respect to the fiber dispersion parameter and the longitudinal nonlinear parameter distribution based on the automatic differentiation mechanism, and iteratively update the fiber dispersion parameter and the longitudinal nonlinear parameter distribution until the error objective function converges.

[0042] Specifically, in step 501, a differentiable scalar error objective function can be constructed based on the fitting degree between the predicted observed signal and the actual observed signal, which serves as the core basis for parameter optimization.

[0043] In step 502, the automatic differentiation mechanism based on the deep learning framework is used to calculate the gradient of the error objective function with respect to the fiber dispersion parameter and the longitudinal nonlinear parameter distribution, which can reduce the difficulty of model implementation to a certain extent. Based on the calculated gradient, the two parameters are iteratively updated through the optimization algorithm. After each iteration, forward propagation is re-executed to calculate the new predicted observation signal and error objective function. When the error objective function drops to a preset threshold or the decrease in multiple consecutive iterations is less than a preset value, the model is determined to have converged, the iteration is stopped, and the current longitudinal nonlinear parameter distribution is output.

[0044] In an optional embodiment, during the iteration process, legal boundary constraints that conform to the physical characteristics of single-mode fiber can be set for the distribution of fiber dispersion parameters and longitudinal nonlinear parameters. During iterative optimization, parameters are updated only within these boundary constraints, avoiding physically meaningless parameter outliers and further improving the robustness of the model. For example, the fiber dispersion parameter β2 of conventional G.652 single-mode fiber in the 1550nm band ranges from -20 to -22 ps² / km, and the optimization boundary of β2 can be constrained within this range.

[0045] Therefore, this application embodiment achieves the joint solution of dispersion parameters and longitudinal nonlinear parameters through end-to-end error optimization, eliminating the need to manually design complex parameter inversion rules, demonstrating strong adaptability, greatly simplifying the implementation process of multi-parameter optimization, while ensuring the accuracy of gradient calculation. Using the convergence of the error objective function as the iteration termination condition ensures the matching degree between the final parameters and the actual link state, providing a reliable parameter basis for subsequent anomaly analysis.

[0046] In one specific embodiment, the formula for determining the error objective function includes: ; in, A [ L [Length is] L The actual observed signal of the aforementioned fiber optic link, A pred [ L ] represents the predicted observation signal output by the differentiable physical model, and γ′ represents the distribution of the longitudinal nonlinear parameters. R (γ′) is a regularization term for the longitudinal nonlinear parameter distribution.

[0047] Specifically, the error objective function consists of two parts: The first part is the reconstruction error term: The mean square error between the predicted and actual observed signals is used to measure the degree of fit between the model-reconstructed signal and the actual received signal. The smaller the value of this item, the closer the signal predicted by the model is to the actual transmission result of the link, and the higher the accuracy of parameter estimation.

[0048] The second part is the regularization term: R(γ′), which is a regularization constraint for the distribution of longitudinal nonlinear parameters. In this embodiment, L1 regularization, L2 regularization or elastic net regularization can be used. Its core function is to constrain the range of values ​​of the nonlinear weight vector, avoid the model from overfitting the noise in the received signal, prevent the occurrence of parameter outliers without physical meaning, and improve the generalization ability of the model.

[0049] In an optional embodiment, the weight coefficient of the regularization term can be adjusted according to the actual signal-to-noise ratio of the optical fiber link. In high signal-to-noise ratio scenarios, the regularization weight can be reduced to improve the parameter estimation accuracy, while in low signal-to-noise ratio scenarios, the regularization weight can be increased to enhance the model's noise resistance.

[0050] In an optional embodiment of this application, the iterative update of the fiber dispersion parameters and the longitudinal nonlinear parameter distributions includes: The distributions of the fiber dispersion parameters and the longitudinal nonlinear parameters are jointly iteratively updated based on the Adam optimization algorithm.

[0051] Specifically, Adam (Adaptive Moment Estimation) is a first-order gradient descent optimization algorithm that can adaptively adjust the learning rate based on the first and second moments of the gradient.

[0052] In an optional embodiment, this application adopts a phased joint iterative update strategy. Specifically, the longitudinal nonlinear parameter distribution is first fixed as an initial value of all zeros, and the fiber dispersion parameter is pre-optimized by the Adam optimization algorithm for a preset fixed number of steps. When the fiber dispersion parameter converges to a preset range that conforms to the physical characteristics of single-mode fiber, the fiber dispersion parameter and the longitudinal nonlinear parameter distribution are then jointly optimized alternately by the Adam optimization algorithm until the error objective function converges to a preset threshold.

[0053] Reference Figure 3 The diagram shown is a convergence diagram of the fiber dispersion parameter β2 during the iterative optimization process in an embodiment of this application. It can be seen that the β2 value of the variable to be optimized gradually approaches a fixed value during the iteration process.

[0054] Based on the Adam optimization algorithm, the convergence speed of parameter iteration can be significantly improved. The phased preheating and alternating optimization strategy solves the optimization instability problem caused by the strong coupling between the fiber dispersion parameters and the longitudinal nonlinear parameters, avoids the error caused by the forced fitting of nonlinear parameters to dispersion mismatch, ensures the estimation accuracy of the two parameters, and further improves the stability of the scheme.

[0055] In an alternative embodiment, during the preheating phase, the problem of finding a local optimum can be solved by using randomized multi-starting-point dispersion initial values.

[0056] In an optional embodiment of this application, the construction of a differentiable physical model includes: A linear dispersion calculation branch is constructed, which is used to calculate the linear ideal optical field of the modulated optical signal at the receiving end of the optical fiber link based on the frequency domain dispersion transmission operator, and to calculate the piecewise linear optical field of the modulated optical signal at each longitudinal segment position of the optical fiber link.

[0057] Reference Figure 4 As shown, the linear dispersion calculation branch is the basic branch of the model. It achieves efficient calculation of linear dispersion through FFT (Fast Fourier Transform) and IFFT (Inverse Fast Fourier Transform). This branch has two outputs: one is the linear ideal optical field of the modulated optical signal transmitted to the receiving end when only considering the linear dispersion without nonlinear effects; the other is the piecewise linear optical field of the modulated optical signal transmitted to each longitudinal segment position of the fiber optic link, which provides input for subsequent nonlinear perturbation calculation.

[0058] A nonlinear perturbation calculation branch is constructed, which is used to calculate the nonlinear perturbation kernel coefficients corresponding to each segment based on the piecewise linear optical field, so as to construct a nonlinear perturbation matrix according to the nonlinear perturbation kernel coefficients.

[0059] Reference Figure 4 As shown, the nonlinear disturbance calculation branch is the correction branch of the model. Based on the first-order perturbation theory, the nonlinear effect of the optical fiber link can be regarded as a small perturbation on the basis of linear transmission. This branch calculates the nonlinear disturbance kernel coefficients generated by each segment and transmitted to the receiving end based on the piecewise linear optical field, and then splices the nonlinear disturbance kernel coefficients of all segments into a nonlinear disturbance matrix, which fully characterizes the nonlinear disturbance characteristics of the entire optical fiber link.

[0060] Based on the superposition relationship between the linear ideal optical field and the nonlinear perturbation matrix, a mapping function is established between the predicted observation signal and the distribution of the fiber dispersion parameter and the longitudinal nonlinear parameter.

[0061] Reference Figure 4 As shown, based on the linear and nonlinear superposition characteristics of optical fiber transmission, the weighted superposition result of the linear ideal optical field and the nonlinear perturbation matrix is ​​used as the prediction observation signal. A mapping function between the prediction observation signal and the two parameters to be optimized is established, thereby completing the construction of the differentiable physical model.

[0062] In an optional embodiment, the formula for calculating the linear ideal light field includes: ; The calculation formula for the piecewise linear optical field includes: ; The formula for calculating the nonlinear perturbation kernel coefficients includes: ; The formula for calculating the nonlinear perturbation matrix includes: ; The formula for determining the mapping function includes: ; Where, A0[ [ ] represents the linear ideal light field. L The total transmission length of the optical fiber link is given by A0[L], which is the linear ideal optical field of the optical fiber link of length L. A[0] is the modulated optical signal, β2 is the optical fiber dispersion parameter, ω is the angular frequency, and n0[ ] represents the piecewise linear optical field. Let z be the distance from the transmitter of the optical fiber link to the k-th longitudinal segment position. k The transmission length, n0[ ], that is, the transmission length is Piecewise linear optical field, FFT( ) is the Fast Fourier Transform, IFFT( ) represents the inverse fast Fourier transform, and Δz represents the unit step size of the longitudinal segmentation of the optical fiber link. The nonlinear perturbation kernel coefficients are... For from the first k The vertical segment position z k The frequency domain dispersion transmission operator to the receiving end of the optical fiber link, That is, the result used to calculate the nonlinear disturbances generated by segmentation after transmission through the remaining fiber length. N ( ) is a nonlinear mapping operator. N ( This involves performing nonlinear mapping calculations on piecewise linear light fields, specifically... N ( )= , For the squared magnitude of the piecewise linear optical field, the corresponding optical power at the segment location is... Let m be the nonlinear perturbation matrix. z γ is the total number of segments in the optical fiber link; γ′ is the longitudinal nonlinear parameter distribution. That is, the matrix product of the nonlinear perturbation matrix and the longitudinal nonlinear parameter distribution, representing the total nonlinear perturbation generated by all segments of the fiber optic link, and is a nonlinear correction term for the linear ideal optical field, A. pred [ ] represents the predicted observation signal. This refers to the predicted observation signal of an optical fiber link of length L, where j is the imaginary unit.

[0063] The differentiable physical model in this embodiment is constructed based on the first-order perturbation theory of optical fiber transmission. All computational steps have clear physical meanings, which is different from black-box neural networks without physical constraints. It is interpretable and avoids the problems of poor generalization and uncontrollable results of AI models. It adopts the frequency domain dispersion calculation method, which can improve the computational efficiency compared with the time domain segment-by-segment calculation. It is easy to implement in engineering and can be adapted to application scenarios with high real-time requirements.

[0064] Reference Figure 5 As shown, in an optional embodiment of this application, the step of performing anomaly analysis on the optical fiber link based on the converged longitudinal nonlinear parameter distribution includes: Step 601: Map the converged longitudinal nonlinear parameter distribution to the optical power value at each longitudinal position.

[0065] Based on the physical laws of the Kerr nonlinear effect in optical fiber, the intensity of the nonlinear effect is strictly linearly proportional to the optical power. Therefore, each component of the converged longitudinal nonlinear parameter distribution has a linear mapping relationship with the average optical power value of the corresponding optical fiber link segment. Through this mapping relationship, the nonlinear weight component can be deduced into the optical power value of the corresponding segment.

[0066] Step 602: Generate the longitudinal optical power curve of the optical fiber link based on the optical power value.

[0067] Arrange all segmented optical power values ​​in order of position along the fiber transmission direction, and generate a continuous longitudinal optical power curve through smooth fitting, referring to... Figure 6 As shown, the horizontal axis of the curve represents the vertical transmission length of the fiber optic link (…). Figure 6 The vertical axis represents the optical power value at the corresponding location (transmission distance). Figure 6 For medium-power, normal, and fault-free fiber optic links, the curve exhibits a smooth exponential attenuation trend consistent with the inherent attenuation characteristics of optical fibers.

[0068] Step 603: Identify the local abrupt change locations in the longitudinal optical power curve to determine the abnormal loss locations, and determine the loss amplitude of the optical fiber link based on the power difference before and after the local abrupt change locations.

[0069] When anomalies occur in the optical fiber link, such as splice degradation, bending and squeezing, connector failure or optical cable damage, the optical power at the abnormal location will drop abruptly in a localized manner. By identifying the local abrupt change characteristics in the longitudinal optical power curve, the horizontal coordinate corresponding to the abrupt change point is the location where the abnormal loss occurs, and the difference in optical power before and after the abrupt change is the additional loss amplitude at that abnormal point.

[0070] Therefore, this embodiment can achieve online longitudinal power monitoring and fault diagnosis of fiber optic links without interrupting services and without additional hardware costs. It also eliminates the need for offline deployment of testing equipment such as OTDRs, significantly reducing the operation and maintenance costs of fiber optic links. Based on the accurate parameter inversion of converged optical power distribution, the influence of dispersion mismatch is eliminated, the positioning accuracy of abnormal loss is improved, the loss amplitude error is smaller, and minute anomalies in the link can be accurately identified, enabling early warning of faults. The complete anomaly analysis process can be automated without manual intervention and can be directly integrated into the operation and maintenance management platform of the fiber optic communication system to achieve real-time monitoring and intelligent diagnosis of link status.

[0071] In an optional embodiment, local abrupt changes in the longitudinal optical power curve can be identified by a sliding window difference algorithm, or anomalies can be determined by a preset power abrupt change threshold; no absolute limitation is imposed here.

[0072] It is understandable that the steps in the above method are only for clear description. In implementation, they can be combined into one step or some steps can be split into multiple steps. As long as they include the same logical relationship, they are all within the scope of protection of this application. Adding insignificant modifications or introducing insignificant designs to the algorithm or process, but without changing the core design of the algorithm and process, are all within the scope of protection of this patent.

[0073] Furthermore, the examples mentioned in the above embodiments can be freely combined, and any combination can be understood as an embodiment. The terms "embodiment" or "example" appearing in various locations in the specification do not necessarily refer to the same embodiment, nor are they independent or alternative embodiments mutually exclusive with other embodiments. Those skilled in the art will understand that the embodiments described herein can be combined with other embodiments.

[0074] Another embodiment of this application relates to an electronic device, see reference Figure 7 As shown, it includes at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method described above.

[0075] The memory and processor are connected via a bus, which can include any number of interconnecting buses and bridges, connecting various circuits of one or more processors and memories. The bus can also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and will not be described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over the wireless medium via an antenna, which further receives data and transmits it to the processor.

[0076] The processor manages the bus and general processing, and also provides various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory is used to store data used by the processor during operation.

[0077] Another embodiment of this application relates to a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the method embodiments described above.

[0078] That is, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0079] Those skilled in the art will understand that the above embodiments are specific embodiments for implementing this application, and in practical applications, various changes can be made to them in form and detail without departing from the spirit and scope of this application.

Claims

1. A method for analyzing fiber optic link anomalies, characterized in that, include: Acquire the modulated optical signal from the transmitting end of the optical fiber link, and the actual sampled signal from the receiving end of the optical fiber link; The actual observation signal is generated based on the actual sampled signal; A differentiable physical model is constructed, wherein the parameters to be optimized in the differentiable physical model include the fiber dispersion parameters and longitudinal nonlinear parameter distributions of the fiber link, and the output of the differentiable physical model is the predicted observation signal corresponding to the actual observation signal. The modulated optical signal is input into the differentiable physical model to generate the predicted observation signal; Based on the predicted observation signal and the actual observation signal, the distribution of the fiber dispersion parameter and the longitudinal nonlinear parameter in the differentiable physical model is iteratively optimized until the differentiable physical model converges; Anomaly analysis is performed on the optical fiber link based on the converged longitudinal nonlinear parameter distribution.

2. The fiber optic link anomaly analysis method according to claim 1, characterized in that, The fiber dispersion parameter is the second-order group velocity dispersion coefficient of the fiber link; the longitudinal nonlinear parameter distribution is a nonlinear weight coefficient vector that is set in equally spaced segments along the transmission direction of the fiber link, wherein each component of the nonlinear weight coefficient vector corresponds one-to-one with a longitudinal segment of the fiber link.

3. The fiber optic link anomaly analysis method according to claim 1, characterized in that, The step of iteratively optimizing the distributions of the fiber dispersion parameters and the longitudinal nonlinear parameters in the differentiable physical model based on the predicted observation signals and the actual observation signals until the differentiable physical model converges includes: Construct an objective function for the error between the predicted observed signal and the actual observed signal; The gradient of the error objective function with respect to the fiber dispersion parameter and the longitudinal nonlinear parameter distribution is calculated based on the automatic differentiation mechanism. The fiber dispersion parameter and the longitudinal nonlinear parameter distribution are iteratively updated until the error objective function converges.

4. The fiber optic link anomaly analysis method according to claim 3, characterized in that, The formula for determining the objective function of the error includes: ； in, A [ L [Length is] L The actual observed signal of the aforementioned fiber optic link, A pred [ L ] represents the predicted observation signal output by the differentiable physical model, and γ′ represents the distribution of the longitudinal nonlinear parameters. R (γ′) is a regularization term for the longitudinal nonlinear parameter distribution.

5. The fiber optic link anomaly analysis method according to claim 3, characterized in that, The iterative update of the fiber dispersion parameters and the longitudinal nonlinear parameter distributions includes: The distributions of the fiber dispersion parameters and the longitudinal nonlinear parameters are jointly iteratively updated based on the Adam optimization algorithm.

6. The fiber optic link anomaly analysis method according to claim 1, characterized in that, The construction of the differentiable physical model includes: A linear dispersion calculation branch is constructed, which is used to calculate the linear ideal optical field of the modulated optical signal at the receiving end of the optical fiber link based on the frequency domain dispersion transmission operator, and to calculate the piecewise linear optical field of the modulated optical signal at each longitudinal segment position of the optical fiber link. A nonlinear perturbation calculation branch is constructed, which is used to calculate the nonlinear perturbation kernel coefficients corresponding to each segment based on the piecewise linear optical field, so as to construct a nonlinear perturbation matrix according to the nonlinear perturbation kernel coefficients; Based on the superposition relationship between the linear ideal optical field and the nonlinear perturbation matrix, a mapping function is established between the predicted observation signal and the distribution of the fiber dispersion parameter and the longitudinal nonlinear parameter.

7. The fiber optic link anomaly analysis method according to claim 6, characterized in that, The formula for calculating the linear ideal light field includes: ； The calculation formula for the piecewise linear optical field includes: ； The formula for calculating the nonlinear perturbation kernel coefficients includes: ； The formula for calculating the nonlinear perturbation matrix includes: ； The formula for determining the mapping function includes: ； Among them, A0[ [This refers to the linear ideal light field.] L The total transmission length of the optical fiber link is given, A[0] is the modulated optical signal, β2 is the optical fiber dispersion parameter, ω is the angular frequency, and n0[ ] represents the piecewise linear optical field. Let z be the distance from the transmitter of the optical fiber link to the k-th longitudinal segment position. k The transmission length, FFT( ) is the Fast Fourier Transform, IFFT( ) represents the inverse fast Fourier transform, and Δz represents the unit step size of the longitudinal segmentation of the optical fiber link. The nonlinear perturbation kernel coefficients are... For from the first k The vertical segment position z k The frequency domain dispersion transmission operator to the receiving end of the optical fiber link, N ( ) is a nonlinear mapping operator. Let m be the nonlinear perturbation matrix. z γ is the total number of segments in the optical fiber link; γ′ is the longitudinal nonlinear parameter distribution, A pred [ ] represents the predicted observation signal.

8. The fiber optic link anomaly analysis method according to any one of claims 1-7, characterized in that, The anomaly analysis of the optical fiber link based on the converged longitudinal nonlinear parameter distribution includes: The converged longitudinal nonlinear parameter distribution is mapped to the optical power value at each longitudinal position; The longitudinal optical power curve of the optical fiber link is generated based on the optical power value; Identify the locations of local abrupt changes in the longitudinal optical power curve to determine the locations of abnormal losses, and determine the loss amplitude of the optical fiber link based on the power difference before and after the locations of the local abrupt changes.

9. The fiber optic link anomaly analysis method according to claim 1, characterized in that, The step of generating the actual observation signal based on the actual sampled signal includes: The actual sampled signal is sequentially resampled, dispersion pre-compensated, polarization demultiplexed, and carrier phase recovered to obtain the baseband processed signal; The baseband processed signal is reloaded with the compensated linear dispersion to generate the actual observed signal.

10. An electronic device, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method as described in any one of claims 1 to 9.

Images