Fiber Optic Anomaly Detection Methods and Equipment

Abstract

This application provides a method and device for optical fiber anomaly detection, relating to the field of transmission technology. The method includes: acquiring the transmit power spectrum of the target optical fiber link under a reference state and the receive power spectrum monitored online; determining the abnormal insertion loss value based on the two total powers; employing a power evolution model that includes the stimulated Raman scattering effect between channels, evolving the transmit power spectrum forward to obtain a forward spectral tilt distribution, and evolving the receive power spectrum backward to obtain a reverse spectral tilt distribution; utilizing the matching characteristic of the forward and reverse spectral tilt distributions at the anomaly location, deriving a closed-form solution calculation formula for optical fiber anomalies to accurately determine the location of the anomaly. This application utilizes only the transmit and receive power spectra of existing transmission channels to achieve fast and accurate optical fiber anomaly localization and quantification through a closed-form solution, requiring no additional hardware, avoiding complex calculations and large-scale data training, and possessing advantages such as strong physical interpretability, good real-time performance, and suitability for low-cost distributed deployment.

Description

Technical Field

[0001] This application relates to the field of transmission technology, and in particular to methods and equipment for detecting optical fiber anomalies. Background Technology

[0002] With the rapid development of technologies such as cloud computing and artificial intelligence, global data traffic has surged. As a core transmission infrastructure, the stable and reliable operation of optical networks is crucial. Currently, optical networks are evolving from the traditional C-band to multi-band systems to meet greater capacity demands. However, in multi-band systems, stimulated Raman scattering (SRS) is significantly enhanced, leading to a power spectrum tilt and making system performance more sensitive to anomalies in fiber optic links (such as bends, breaks, or connection losses). These anomalies can introduce additional insertion loss, degrade signal quality, and even cause service interruptions. Therefore, developing a technology capable of rapidly and accurately detecting the location and magnitude of fiber optic anomalies in multi-band environments is key to ensuring the reliable operation and maintenance of optical networks.

[0003] Currently, there are three main technical approaches for fiber optic anomaly detection. The first is a hardware-based detection scheme using an optical time-domain reflectometer (OTDR), which locates anomalies by emitting probe light pulses into the fiber and analyzing the backscattered signal. The second is a data-driven neural network-based scheme, which achieves intelligent identification by training a model to learn the mapping relationship between optical performance indicators and anomalies. The third is a digital signal processing-based scheme, which detects anomalies by matching the actual received signal with the simulated signal to invert the link power distribution.

[0004] However, the aforementioned existing solutions all have significant limitations and cannot meet the demands of modern optical networks for low cost, high real-time performance, strong interpretability, and easy distributed deployment. Specifically, hardware detection solutions require dedicated equipment, are costly, and have slow response times; neural network solutions rely on large amounts of hard-to-obtain anomaly data, have poor model generalization ability, and lack physical interpretability; digital signal processing solutions are computationally complex, have large storage requirements, and lack real-time performance and deployment flexibility. Therefore, there is an urgent need for a fiber optic anomaly detection method that does not rely on additional hardware, requires no large amount of training data, is computationally efficient, and has a clear mechanism. Summary of the Invention

[0005] In view of this, embodiments of this application provide a method and apparatus for detecting optical fiber anomalies, in order to eliminate or improve one or more defects existing in the prior art.

[0006] One aspect of this application provides a method for detecting fiber optic anomalies, comprising:

[0007] Obtain the transmit power spectrum of the target fiber optic link under reference conditions, and the receive power spectrum obtained through online monitoring;

[0008] Based on the total power of the transmitted power spectrum and the total power of the received power spectrum, the abnormal insertion loss value of the target optical fiber link is determined;

[0009] Based on the power evolution model that includes the inter-channel stimulated Raman scattering (ISRS) effect, the transmit power spectrum is calculated in the forward direction to obtain the forward spectrum tilt distribution, and the receive power spectrum is calculated in the reverse direction to obtain the reverse spectrum tilt distribution.

[0010] Based on the characteristic that the forward spectral tilt distribution and the reverse spectral tilt distribution match at the fiber optic anomaly location, the location of the anomaly in the target fiber optic link is determined.

[0011] In some embodiments of this application, before obtaining the transmit power spectrum of the target optical fiber link under reference conditions, the method further includes:

[0012] The key parameters in the power evolution model are optimized using training data generated from the digital twin model of the target fiber optic link, and the power evolution model is configured with the optimized key parameters; wherein, the key parameters include: Raman gain slope, reference frequency, and equivalent attenuation index.

[0013] In some embodiments of this application, optimizing key parameters in the power evolution model using training data generated from the digital twin model of the target fiber optic link, and configuring the power evolution model with the optimized key parameters, includes:

[0014] The digital twin model is constructed based on the physical parameters of the target fiber optic link; wherein, the physical parameters include at least one of fiber length, attenuation coefficient, nonlinear coefficient, effective area, and Raman gain spectrum;

[0015] Using the digital twin model, multiple anomalous events are simulated to generate multiple sets of training data; wherein each anomalous event is defined by an anomalous location and an anomalous size, and each set of training data includes a corresponding simulated transmit power spectrum, simulated anomalous receive power spectrum, anomalous location label, and annomalous size label;

[0016] Based on the power evolution model that includes the inter-channel stimulated Raman scattering (ISRS) effect, a parameter set containing the aforementioned key parameters is determined;

[0017] The parameter set is iteratively optimized to obtain an optimized parameter set with the goal of minimizing the loss function value. The loss function value represents the difference between the predicted power spectrum calculated by the power evolution model based on the simulated transmit power spectrum under the current value of the parameter set and the corresponding simulated abnormal receive power spectrum, as well as the difference between the spectral tilt of the predicted power spectrum and the spectral tilt of the corresponding simulated abnormal receive power spectrum.

[0018] Configure the power evolution model based on the optimized parameter set.

[0019] In some embodiments of this application, determining the abnormal insertion loss value of the target optical fiber link based on the total power of the transmitted power spectrum and the total power of the received power spectrum includes:

[0020] Calculate a first difference between the total power of the transmitted power spectrum and the total power of the received power spectrum; wherein the first difference is a value in decibels;

[0021] Based on the optimized equivalent attenuation index, the inherent attenuation of the target optical fiber link is calculated; wherein, the inherent attenuation is the power attenuation value caused by the inherent characteristics of the optical fiber when the total power of the transmitted power spectrum is transmitted to the measurement position of the received power spectrum under the normal state of the target optical fiber link.

[0022] The abnormal insertion loss value is determined based on the second difference between the first difference and the inherent attenuation.

[0023] In some embodiments of this application, the step of performing forward evolution calculations on the transmit power spectrum to obtain a forward spectral tilt distribution based on a power evolution model that includes the inter-channel stimulated Raman scattering (ISRS) effect, and performing reverse evolution calculations on the receive power spectrum to obtain a reverse spectral tilt distribution, includes:

[0024] A linear regression is performed on the logarithm of the signal power in the forward power distribution with frequency to calculate the forward spectral tilt distribution; wherein, the forward spectral tilt distribution characterizes: the regression slope of the logarithm of the signal power in the forward power distribution as a function of frequency at any position along the signal transmission direction;

[0025] The forward spectrum tilt distribution is calculated based on the forward power distribution; wherein, the forward spectrum tilt distribution characterizes the degree of tilt of the logarithm of the signal power in the forward power distribution as a function of frequency at any position along the signal transmission direction;

[0026] Furthermore, based on the power evolution model and the received power spectrum, an evolution calculation is performed in the opposite direction to the signal transmission to obtain a reverse power distribution; wherein, the reverse power distribution represents: the estimated signal power of each frequency channel at any position along the signal transmission direction in the target optical fiber link, starting from the received power spectrum;

[0027] A linear regression is performed on the logarithm of the signal power in the reverse power distribution with frequency to calculate the reverse spectrum tilt distribution; wherein, the reverse spectrum tilt distribution characterizes the regression slope of the logarithm of the signal power in the reverse power distribution with frequency at any position along the signal transmission direction.

[0028] In some embodiments of this application, determining the location of the anomaly in the target fiber optic link based on the characteristic that the forward spectral tilt distribution and the reverse spectral tilt distribution match at the fiber optic anomaly location includes:

[0029] Based on the forward spectral tilt distribution and the reverse spectral tilt distribution, the location of the anomaly in the target optical fiber link is calculated using a closed-form solution constructed based on the matching condition of equal spectral tilt.

[0030] In some embodiments of this application, the step of iteratively optimizing the parameter set to obtain an optimized parameter set with the goal of minimizing the loss function value includes:

[0031] The Particle Swarm Optimization (PSO) algorithm is used to perform at least one iteration of optimization steps on the parameter set until the preset optimization termination condition is met, and the parameter values ​​in the parameter set after the termination iteration are used as the optimized parameter set.

[0032] The optimization steps include:

[0033] Based on the initial parameter values ​​of the parameter set in the current iteration round, the power evolution model is used to calculate the simulated transmit power spectrum to obtain the predicted power spectrum for the current iteration round.

[0034] Based on the predicted power spectrum of the current iteration round and the corresponding simulated abnormal received power spectrum, calculate the loss function value of the current iteration round;

[0035] Based on the loss function value of the current iteration, the parameter values ​​of the parameter set are updated according to the update rules of the Particle Swarm Optimization (PSO) algorithm, so as to serve as the initial parameter values ​​in the next iteration.

[0036] In some embodiments of this application, calculating the loss function value for the current iteration based on the predicted power spectrum of the current iteration and the corresponding simulated anomalous received power spectrum includes:

[0037] Calculate a first error term and a second error term; wherein the first error term represents the mean square error between the predicted power spectrum of the current iteration round and the corresponding simulated abnormal received power spectrum; and the second error term represents the error between the spectral tilt of the predicted power spectrum of the current iteration round and the spectral tilt of the corresponding simulated abnormal received power spectrum.

[0038] The loss function value for the current iteration is determined by the weighted sum of the first error term and the second error term.

[0039] A second aspect of this application provides an optical fiber anomaly detection device, comprising:

[0040] The power acquisition module is used to acquire the transmit power spectrum of the target fiber optic link under reference conditions, as well as the receive power spectrum obtained through online monitoring.

[0041] An anomaly size determination module is used to determine the abnormal insertion loss value of the target optical fiber link based on the total power of the transmitted power spectrum and the total power of the received power spectrum;

[0042] The anomaly localization module is used to perform forward evolution calculation on the transmit power spectrum based on a power evolution model that includes the inter-channel stimulated Raman scattering (ISRS) effect, to obtain a forward spectrum tilt distribution, and to perform reverse evolution calculation on the receive power spectrum, to obtain a reverse spectrum tilt distribution; based on the characteristic that the forward spectrum tilt distribution and the reverse spectrum tilt distribution match the anomaly location in the optical fiber, the anomaly location of the target optical fiber link is determined.

[0043] A third aspect of this application provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the fiber optic anomaly detection method.

[0044] A fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the described fiber optic anomaly detection method.

[0045] The fifth aspect of this application provides a computer program product comprising a computer program that, when executed by a processor, implements the aforementioned fiber optic anomaly detection method.

[0046] The fiber optic anomaly detection method provided in this application uses only the transmit and receive power spectra at both ends of the target fiber optic link as input data. This solves the high cost and deployment flexibility issues associated with traditional optical time domain reflectometer (OTDR) schemes, which require additional dedicated hardware. Furthermore, it eliminates the need for additional hardware overhead and is easy to integrate and deploy in a distributed manner. By determining the anomaly loss value through closed-form calculation based on the total power of the power spectrum, it addresses the dependence of data-driven methods such as neural networks on massive amounts of anomaly training data, effectively reducing data requirements, avoiding model black-box architecture, and improving applicability. Finally, by employing a physical power evolution model incorporating inter-channel stimulated Raman scattering (ISRS) effects for calculating the forward and reverse spectral tilt distribution, it addresses the challenges of digital signal processing (DSP)... Numerical inversion methods such as Processing and Digital Spinning (DSP) are computationally complex and require large storage, making the computation process simple, efficient, and physically interpretable. By locating anomalies based on the matching characteristics of spectral tilt distribution at the location of anomalies, it can solve the problem that existing methods cannot simultaneously achieve real-time detection, positioning accuracy, and mechanistic clarity, thus enabling fast, accurate, and mechanistically transparent fiber optic anomaly localization and diagnosis.

[0047] Additional advantages, objectives, and features of this application will be set forth in part in the description which follows, and will in part become apparent to those skilled in the art upon review of the following description, or may be learned by practice of the application. The objectives and other advantages of this application can be realized and obtained by means of the structures specifically pointed out in the specification and drawings.

[0048] Those skilled in the art will understand that the purposes and advantages that can be achieved with this application are not limited to those specifically described above, and that the above and other purposes that this application can achieve will be more clearly understood from the following detailed description. Attached Figure Description

[0049] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, do not constitute a limitation thereof. The components in the drawings are not drawn to scale but are merely for illustrating the principles of this application. For ease of illustration and description of certain parts of this application, corresponding portions in the drawings may be enlarged, i.e., may appear larger relative to other components in an exemplary device actually manufactured according to this application. In the drawings:

[0050] Figure 1 The diagram shows the changes in SRS performance and the distribution of received power caused by different abnormal attenuations occurring at different locations, as provided in the basic principles of this application.

[0051] Figure 2 This is a schematic diagram illustrating the invariance of spectral tilt when an optical fiber anomaly occurs, as provided in the basic principles of this application.

[0052] Figure 3 This is a schematic diagram of the first process of the fiber optic anomaly detection method in one embodiment of this application.

[0053] Figure 4 This is a schematic diagram of a second process of the fiber optic anomaly detection method in one embodiment of this application.

[0054] Figure 5 This is a schematic diagram of the specific process of step 010 in the optical fiber anomaly detection method in one embodiment of this application.

[0055] Figure 6 This is a flowchart illustrating the fiber optic anomaly detection method in an application example of this application.

[0056] Figure 7(a) is a schematic diagram comparing the calculation error of the anomaly size in the closed-form solution calculation results after PSO search for optimal parameters in an application example of this application.

[0057] Figure 7(b) is a schematic diagram comparing the calculation errors of abnormal positions in the closed-form solution calculation results after PSO search for optimal parameters in an application example of this application.

[0058] Figure 8 This is a schematic diagram of the fiber optic anomaly detection device in one embodiment of this application. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and their descriptions are used to explain this application, but are not intended to limit it.

[0060] It should also be noted that, in order to avoid obscuring this application with unnecessary details, only the structures and / or processing steps closely related to the solution according to this application are shown in the accompanying drawings, while other details that are not closely related to this application are omitted.

[0061] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, element, step, or component, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0062] It should also be noted that, unless otherwise specified, the term "connection" in this article can refer not only to a direct connection, but also to an indirect connection involving an intermediary.

[0063] In the following description, embodiments of the present application will be illustrated with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar parts, or the same or similar steps.

[0064] First, it's important to clarify that optical networks, as the infrastructure for large-scale information transmission, play a crucial role in data transmission and service assurance in emerging technologies such as computing power interconnection, large-scale distributed inference training, and digital twin data telemetry and real-time updates. These services place higher demands on the network, including greater bandwidth, lower latency, and higher quality of service, while also requiring more sophisticated intelligent and stable operation and maintenance, accurate fault monitoring, and automated network repair. Throughout the optical network's lifecycle, optical fibers are subject to interference from external activities such as road construction, pipeline maintenance, and natural disasters, leading to frequent fiber anomalies. These anomalies manifest as persistent or transient fiber loss anomalies, causing additional insertion loss at certain points in the fiber, ultimately resulting in a decrease in signal-to-noise ratio, interruption of online services, or even serious system failures. Furthermore, to meet the ever-increasing capacity and transmission demands, optical networks are evolving from traditional single-C-band transmission to multi-band transmission. Their more complex system structure and wider coverage make them more closely intertwined with urban, road, and railway networks, introducing more potential anomalies into the lifecycle management process. Meanwhile, in multi-band transmission systems, stimulated Raman scattering (SRS) within optical fibers is significantly enhanced, causing power to shift from high-frequency channels to low-frequency channels, thus weakening the overall transmission performance of the system. This problem is particularly pronounced when network conditions become increasingly dynamic and nonlinear effects become more significant. Therefore, in multi-band optical networks, fiber anomalies are expected to cause more severe damage and generate higher maintenance costs than in traditional C-band transmission systems. Thus, accurately locating and estimating the position and magnitude of fiber anomalies is crucial for ensuring the stable and reliable operation of the optical network throughout its entire lifecycle.

[0065] Currently, there are roughly three methods to detect fiber optic anomalies.

[0066] Option 1: Hardware detection based on an optical time-domain reflectometer (OTDR) is currently the most commonly used anomaly monitoring method for fiber optic lines. Its working principle is as follows: The OTDR emits an out-of-band single-wavelength probe pulse into the fiber under test. As the pulse propagates through the fiber, continuous Rayleigh backscattering occurs due to refractive index perturbations, while Fresnel reflections occur at discontinuous interfaces such as splices, connector ends, breaks, or bends. The OTDR receives and collects these backscattered light signals that return over time using a high-speed detector. The time axis is converted to a distance axis using the group refractive index of light propagating in the fiber, thus obtaining the attenuation distribution curve along the fiber. By analyzing the slope changes, reflection peaks, and power abrupt changes of this curve, anomalies, splice losses, and the attenuation spectrum along the fiber can be detected and located.

[0067] Option 2: Fiber optic anomaly detection method based on a data-driven neural network model. This method leverages the impact of fiber optic anomalies on observable optical performance indicators by constructing a mapping relationship between input and output to achieve intelligent identification of fiber optic status. Specifically, when anomalies such as fiber breakage, bending, micro-bending, poor connection, or external disturbances occur, various optical performance parameters, including Q-factor, bit error rate (BER), received optical power, and state of polarization (SOP), will change. Therefore, this type of method typically collects optical performance indicator data containing multiple anomaly types through simulation or experimentation, and trains a neural network model using supervised learning to learn the equivalent mapping relationship between optical performance indicators and fiber optic anomalies, thereby achieving anomaly detection and localization. For example, some researchers have used bidirectional gated recurrent units (BiGRUs) to construct detection models, achieving the identification and localization of anomalies such as fiber breakage, contaminated connections, poor splices, and fiber optic eavesdropping. Furthermore, since the polarization state of optical signals can also reflect the physical changes after the fiber optic link is disturbed, the authors further adopted a generative adversarial network (GAN) to determine whether there are anomalies in the fiber optic link, and developed a classification and localization model based on a vision transformer (ViT), which can detect, locate and classify ten types of anomalies, including fiber bending, shaking, impact, vertical movement and arbitrary combinations of the above events.

[0068] Option 3: The DSP-based fiber optic anomaly detection scheme essentially utilizes the deterministic relationship between the nonlinear phase rotation and power evolution of optical signals during fiber transmission. By optimally matching the actual received signal acquired by the coherent receiver with the reference received signal generated in the simulated link, the distribution of optical power along the transmission distance can be estimated. Specifically, this method establishes a physical transmission model between the transmitting and receiving signals based on the nonlinear Schrödinger Equation (NLSE), and uses the accumulated Kerr nonlinear phase rotation in the fiber as an indicator of power evolution. The receiver first acquires the actual symbol sequence, then simulates the power evolution curves under different assumptions segment by segment to generate corresponding reference symbol sequences. Subsequently, criteria such as correlation-based methods (CMs), minimum-mean-square-error-based methods (MMSE), or linear least squares (LLS) are used to measure the difference between the actual and reference signals. By minimizing this difference, the power evolution curve that best matches the simulated link with the actual link can be derived. The power evolution curve reflects the attenuation, gain, and local disturbances experienced by the optical signal in each fiber segment. Abrupt points, abnormal slope changes, or local power dips in its spatial distribution can indicate bends, compressions, poor connections, or other physical anomalies at any location in the fiber, enabling anomaly detection of the fiber optic link.

[0069] However, the hardware detection method based on OTDR proposed in Scheme 1 requires the introduction of additional hardware. As the scale of optical networks continues to expand, the hardware detection method will inevitably bring additional detection costs. At the same time, it cannot be deployed in a distributed manner and lacks detection flexibility. In addition, due to the existence of stimulated Raman scattering, the out-of-band probe light pulses emitted by the OTDR may cause the probe signal power to shift in multiple bands, affecting the detection accuracy. In fact, the response time of OTDR is generally in the range of seconds to minutes, which cannot meet the requirements for rapid and accurate location of fiber optic anomalies.

[0070] Option 2, which uses a data-driven neural network approach, typically requires a large amount of training data. However, the amount of real-world anomaly data in the existing network is insufficient to meet the data requirements for neural network training. Furthermore, because different optical links have different configurations and characteristics, a unique neural network model for anomaly detection is needed for each specific link, making it impossible to generalize and universally adapt to various link configurations. Thirdly, although the trained neural network model can detect fiber optic link anomalies based on input features, it remains essentially a black box model, making it impossible for network administrators to understand the neural network's output calculation method and lacking physical interpretability.

[0071] Scheme 3 proposes a DSP-based fiber optic anomaly detection method that does not introduce additional hardware. It relies on the received signal from a coherent receiver to deduce the power evolution curve along the link direction. However, this scheme requires not only substantial storage resources to store the received signal data but also complex calculations to obtain the power evolution curve for fiber anomaly detection. This DSP-based approach, requiring significant storage space and involving complex calculations, is unsuitable for low-cost distributed deployment. Furthermore, its inability to respond to rapid fiber optic flashover anomalies further limits its practical application.

[0072] Based on this, in order to address the problems of existing fiber optic anomaly detection schemes relying on additional hardware, requiring large amounts of training data, being computationally complex and lacking physical interpretability, and failing to balance real-time performance, accuracy, and low-cost distributed deployment, this application provides a fiber optic anomaly detection method, a fiber optic anomaly detection device for executing the method, a physical device, a computer-readable storage medium, and a computer program product. These methods utilize only the transmit and receive power spectra of existing transmission channels to achieve fast and accurate fiber optic anomaly localization and quantification through closed-form solutions. They require no additional hardware, avoid complex calculations and large-scale training, and offer advantages such as strong physical interpretability, good real-time performance, and suitability for low-cost distributed deployment.

[0073] The following examples will provide a detailed description.

[0074] First, the basic principle of the fiber optic anomaly detection method that can be implemented by a fiber optic anomaly detection device, as provided in the embodiments of this application, is introduced:

[0075] In multi-band optical transmission systems, the channel i The gain or loss experienced due to the combined effect of fiber attenuation and inter-channel Raman gain is described by the following differential equation:

[0076] (1)

[0077] in, Indicates the propagation distance along the optical fiber z Timei The power of each channel, Indicates the propagation distance along the optical fiber z Time j The power of each channel, Let f be the attenuation coefficient of the optical fiber at frequency f. The effective area of ​​the optical fiber. This indicates that the frequency difference is Raman gain coefficient at the location, N This represents the number of channels. Equation (1) assumes the Raman gain coefficient. The variation within the bandwidth of each channel is negligible; z represents the propagation distance along the fiber axis, which can be called fiber propagation distance, or simply distance or position.

[0078] For different i For Raman gain coefficient Approximately in the window The slope within the effective bandwidth of the Raman gain is... trigonometric functions, This is the Raman gain slope. Assume the lowest frequency of the transmitted signal. and highest frequency If there are infinitely many equally spaced channels, such that the power of each channel converges to the power spectral density, then equation (1) can be rewritten as:

[0079] (2)

[0080] in, This represents the optical power spectral density at frequency f at a propagation distance z in the optical fiber, only in [ , The value inside is not 0. It is an integral variable that represents all possible frequencies other than the current frequency of interest f when calculating Raman interactions.

[0081] Assuming the input power spectrum has a small tilt and the window If the frequency range of the input power spectrum can be covered sufficiently large, then the Raman contribution term can be approximated as evolving proportionally to the total power. Therefore, equation (2) can be derived as follows:

[0082] (3)

[0083] in, This represents the input optical power spectral density at the transmitter (z=0); Indicates at the reference frequency The shape function value at that location; Indicates the distance of transmission The total power at the location of the integral variable (i.e., the integral of the entire signal power spectrum over all frequencies at the location of the integral variable). This is the reference frequency that is not affected by the tilt caused by ISRS. It is a shape function used to reflect the power situation involved in the ISRS process, when the bandwidth is less than hour, It can be simplified to:

[0084] (4)

[0085] The change in total power is entirely determined by attenuation and power spectrum distribution, indicating that ISRS contributes little to the total power attenuation. Under constant attenuation, the change in total power is determined only by a constant attenuation coefficient. However, since ISRS affects the power spectrum distribution, it has an indirect impact on the longitudinal distribution of total power. Therefore, it is assumed that the total power decays exponentially and is determined by a single equivalent attenuation coefficient. control:

[0086] (5)

[0087] Here, n is an adjustable parameter that is always positive, namely the equivalent decay exponent, which can affect the accuracy of the solution. Represents frequency attenuation coefficient at nth power; This indicates the frequency at the transmitting end (z=0). Optical power spectral density at the location; This represents the total power at the transmitting end (z=0), that is, the total power of the entire transmitted signal power spectrum, which can be simply referred to as the total transmitted power.

[0088] By adjusting the value of the equivalent attenuation exponent n, different equivalent attenuation coefficients can be calculated. This optimizes the fitting accuracy of the closed-form solution (Equation (6)) of the subsequent power evolution model to the attenuation behavior of the real link (i.e. the subsequent target fiber link).

[0089] Substitute the simplified In equation (3), calculate the integral on the right-hand side and convert the power spectral density to the integral. Replacing it with a propagation signal from a finite channel, we can obtain a closed-form solution to equation (1), which is also the forward evolution closed-form solution of the power evolution model:

[0090] (6)

[0091] in, This represents the signal power of the i-th channel at a propagation distance z; This represents the power of the i-th channel at the transmitter (z=0), i.e., the transmit power spectrum; Indicates the i-th channel at its center frequency The fiber attenuation coefficient at that location, i.e., α( ); Represents a mathematical constant.

[0092] Similarly, this application can specify the maximum length of the optical fiber. L The received power at that point is used to perform inverse calculations of ISRS and fiber attenuation, yielding a closed-form solution based on the inverse evolution of the received power:

[0093] (7)

[0094] in, This represents the estimated power of the i-th channel at a propagation distance z, calculated through reverse evolution. This represents the power of the i-th channel at the fiber optic end, receiver (z=L), i.e., the reference received power spectrum (this is the starting point or boundary condition for reverse evolution. Under ideal reference conditions without anomalies, it represents the normal received power spectrum). This indicates the total length of the fiber optic link.

[0095] In the forward power evolution process, the highest frequency channel only experiences power consumption, transferring its energy to all other channels, while the lowest frequency channel only accumulates power from all other channels. Therefore, the spectral tilt calculated in the dBm domain is always negative.

[0096] Based on equation (1), this application uses numerical methods to conduct simulation studies on link anomalies under different conditions. For example... Figure 1 As shown, in a single-span C+L band system with a uniform input power spectrum, this application sets up three cases of fiber anomalies: a 2dB anomaly at 10km, a 6dB anomaly at 10km, and a 2dB anomaly at 40km, and finally records the received power, distribution, and spectral tilt. Figure 1As can be seen, the link without any anomalies exhibits the largest received spectral tilt. However, when three anomalies are present, the received spectral tilt decreases to varying degrees, with the three inflection points corresponding to the anomaly locations. Compared to a 2 dB anomaly at 10 km, a 6 dB anomaly at 10 km leads to greater power loss and weakens power transmission, resulting in a smaller spectral tilt at the end of the span. Compared to a 2 dB anomaly at 40 km, a 2 dB anomaly at 10 km reduces power transmission between channels and leads to a smaller received spectral tilt. This indicates that the location and intensity of the anomaly determine the unique power spectral distribution characteristics and the degree of spectral tilt, and there is a one-to-one mapping relationship between the two. That is, a unique anomaly location and intensity correspond to a unique power distribution and tilt pattern. In optical communication systems, the received power spectrum can comprehensively reflect the various power-related impairments and compensations experienced by the signal power during transmission. In multi-band systems, the received power level mainly reflects the inherent loss or additional insertion loss of the optical fiber, while the degree of power spectral tilt reflects the strength of the SRS effect. When there is abnormal insertion loss in the optical fiber, the additional insertion loss introduced by the abnormality will reduce the optical power, which will weaken the SRS effect and thus cause a change in the power spectrum distribution.

[0097] exist Figure 1 In the diagram, the middle box represents the spectral tilt changes under different anomalies; the boxes to the left of this middle box show the input power in the L-band and C-band, and the two boxes to the right of this middle box show the spectral tilt evolution curves for different anomalies at the same location and the same anomaly at different locations, respectively. At a distance of 10 km from the middle box, there are light blue spots representing loss values ​​(2 dB) and red spots representing loss values ​​(6 dB), which are represented by... The two anomalous events (light blue and red spots) are represented at a distance of 10 km; at a distance of 40 km in the middle box, there is a dark blue spot representing the loss value (2 dB), which is indicated by... This indicates an anomaly at a distance of 40 km (dark blue spot).

[0098] in, Figure 1The black line represents the ideal situation where the fiber optic link is free of any anomalies. At this point, the received power spectrum exhibits the largest spectral tilt, reflecting the most complete accumulation of stimulated Raman scattering (ISRS) effects in a complete, healthy link. The light blue line ("10 km, 2 dB anomaly") represents a 2 dB insertion loss anomaly occurring at a distance of 10 km from the transmitter. Compared to the no-anomaly scenario, the overall received power decreases, and the spectral tilt is significantly reduced. This is because the anomaly causes a sharp drop in signal power at 10 km, weakening the power level that induces the ISRS effect in subsequent transmission segments. The red line ("10 km, 6 dB anomaly") represents a more severe 6 dB insertion loss anomaly occurring at the same distance of 10 km. The received power decreases further, and the reduction in spectral tilt is more significant than with the 2 dB anomaly. This indicates that the greater the anomaly loss, the stronger the weakening effect on the ISRS effect and the final spectral tilt. The dark blue line ("40 km, 2 dB anomaly") represents a 2 dB insertion loss anomaly occurring at a distance of 40 km from the transmitter. Its spectral tilt falls between the 2 dB anomaly at 10 km and the absence of anomalies. This indicates that the closer the anomaly occurs to the receiver, the weaker its disruption of the entire link ISRS accumulation process, and therefore the smaller its impact on the final spectral tilt.

[0099] In addition, Figure 1 In the formula, T(z) = linear fit(P[z, f], f), where T(z) represents the spectral tilt value at the optical fiber propagation distance z; P[z, f] represents the channel power at the propagation distance z and frequency f; f represents the frequency, i.e. the channel frequency, which is the independent variable of the linear fit, and linear fit represents the first-order linear fit.

[0100] As can be seen, when the fiber optic anomaly occurs at the location of the anomaly... This causes a drop in power. β At dB, the spectral tilt remains unchanged before and after the anomaly, such as Figure 2 As shown. Meanwhile, the power changes of both parts of the optical fiber before and after the anomaly occur follow equation (1). Therefore, utilizing this characteristic, this application uses the above-mentioned closed-form solution of forward and reverse power evolution to calculate the spectral tilt, and based on the characteristic that the spectral tilt remains unchanged when an optical fiber anomaly occurs, proposes a closed-form optical fiber anomaly detection method.

[0101] exist Figure 2 The diagram is divided into three main parts from left to right, describing the complete process of a signal traveling from the transmitter, through anomalies, and finally reaching the receiver. Figure 2 The left box in the diagram represents a schematic of the abnormal link and the input power P[0,f]. Figure 2 The middle box in the image represents the channel power and core principles before and after the anomaly, with the light blue box showing... The core principle of this application is that the spectral tilt remains unchanged. This represents the decrease in channel power (in dB) in the high-frequency band (usually L-band) at the instant the anomaly occurs. This represents the decrease in channel power (in dB) in the low-frequency band (usually C-band) at the instant the anomaly occurs. Figure 2 The right box in the image indicates abnormal received power. ,in, This indicates the end of the fiber optic link; the total length of the abnormal link is from 0km to... km; Indicates the location where the anomaly occurred, i.e. A βdB anomaly occurred at km. Figure 2 The red spot in the image can represent the... =The abnormal loss value (6dB) at 10km. The dark blue bar before the red spot represents the distance before the fiber optic link anomaly occurred, and the green bar after the red spot represents the distance after the fiber optic link anomaly occurred.

[0102] Based on the above assumptions, ISRS contributes very little to the total power attenuation. That is, this application can assume that ISRS only causes inter-channel power transfer and does not result in power consumption; the total power loss is controlled solely by the effective attenuation coefficient. Therefore, the total transmit power... and received total power The difference is the superposition of the inherent attenuation of the optical fiber and the optical fiber anomaly occurring at any location. The size of the anomaly can be calculated by closed equation (8):

[0103] (8)

[0104] in, This indicates the power ratio caused by anomalies under linear unit conditions; Indicates the total received power; Indicates the total transmission power; This represents the abnormal insertion loss value in the logarithmic field, i.e., the abnormal size.

[0105] Therefore, for abnormal links, if the size of the abnormality is known... The corresponding power ratio can then be calculated using formula (8). Based on and the i Abnormal received power of each channel at the fiber optic end and receiver (z=L) The closed-form solution for the power inverse evolution applicable to anomalous links can be obtained as follows:

[0106] (9)

[0107] in ， This represents the reverse evolution estimated power of the i-th channel under anomalous conditions at a propagation distance z.

[0108] Based on this, embodiments of this application provide a fiber optic anomaly detection method that can be implemented by a fiber optic anomaly detection device, see [link to relevant documentation]. Figure 3 The fiber optic anomaly detection method specifically includes the following:

[0109] Step 100: Obtain the transmit power spectrum of the target fiber optic link under reference conditions, and the receive power spectrum obtained from online monitoring.

[0110] It is understood that the target fiber optic link refers to a section of fiber optic line to be detected for anomalies. It can be an independent fiber optic span or a part of an end-to-end link that is considered as a whole for detection.

[0111] The reference state refers to the moment when the optical fiber link is in a known healthy working state without any abnormalities, and the power spectrum collected at this time serves as the reference baseline for subsequent comparisons.

[0112] The transmit power spectrum and receive power spectrum are the sets of optical power on each wavelength channel, measured at the transmitting and receiving ends of the optical fiber link, respectively. For example, in a C+L band system, it includes the power values ​​of all channels within that band (such as multiple channels in the range of 191.3 THz to 196.0 THz). Since the link may be normal or abnormal when the power data at the receiving end is acquired in real time, the receive power spectrum may be a normal receive power spectrum or an abnormal receive power spectrum. Therefore, it means a power spectrum collected in real time by the receiving end that may already contain unknown anomalies.

[0113] The online monitoring refers to the real-time or periodic collection of power data by optical channel monitors (OCMs) deployed at both ends of the target optical fiber link during its actual operation.

[0114] Specifically, at the transmitter of the target fiber optic link (e.g., an 80km G.652.D single-mode fiber optic link), the optical performance monitoring unit (OCM) deployed there collects and records the optical power values ​​of all working channels (e.g., 96 channels in total, including C-band and L-band) under the initial health state (i.e., baseline state) of the link, forming a transmit power spectrum. This is the reference transmit power spectrum. At the link receiver, the received optical power values ​​of all channels are monitored and collected online and in real time through the OCM, forming the online received power spectrum. That is, receiving the power spectrum online.

[0115] Step 200: Determine the abnormal insertion loss value of the target optical fiber link based on the total power of the transmitted power spectrum and the total power of the received power spectrum.

[0116] It is understood that the total power refers to the sum of the power of all channels in the entire power spectrum, reflecting the overall power level of the signal. The total power of the transmitted power spectrum can be called the total transmitted power. The total power of the received power spectrum can be referred to as the total received power. .

[0117] The abnormal insertion loss value refers to the additional power loss beyond the inherent attenuation of the fiber caused by fiber anomalies (such as bending or compression), and is usually measured in decibels (dB). The abnormal insertion loss value can be simply referred to as the anomaly magnitude. For example, if calculated... β =3.2dB indicates that the anomaly caused an additional power loss of approximately 3.2 dB in the target fiber optic link.

[0118] In step 200, the power values ​​of all channels in the transmit power spectrum and the receive power spectrum are summed to obtain the total transmit power. and received total power The difference between the two in the decibel domain reflects the total attenuation experienced by the signal. Based on the relationship between total power attenuation and inherent fiber attenuation, the inherent attenuation part determined by the fiber material and length is separated from the total attenuation, and the remaining part is the additional insertion loss value introduced by this abnormal event (as shown in Equation (8)).

[0119] Step 300: Based on the power evolution model including the inter-channel stimulated Raman scattering (ISRS) effect, perform forward evolution calculation on the transmit power spectrum to obtain the forward spectrum tilt distribution, and perform reverse evolution calculation on the receive power spectrum to obtain the reverse spectrum tilt distribution.

[0120] Step 300 is executed after step 100, and can be executed in parallel or sequentially with step 200, depending on the specific application requirements.

[0121] It should be noted that the power evolution model is a physical and mathematical model used to describe the evolution of optical signal power in multi-band optical fiber with transmission distance. Its core is the transmission equation (as shown in Equation (1)) that includes the nonlinear effect of inter-channel stimulated Raman scattering (ISRS) or its simplified closed-form solution (as shown in Equation (6)) obtained after a series of approximations (as shown in Equations (2)-(5)). This model is the basis for the theoretical calculation of this method. A closed-form solution means that the model can be directly substituted with parameters and boundary conditions for calculation without the need for complex iterative analytical formulas.

[0122] The forward evolution calculation uses the transmitted power spectrum as the initial condition and uses the model to simulate the evolution of optical power during the transmission process from the transmitter to the receiver (as shown in Equation (6)); the reverse evolution calculation uses the received power spectrum as the endpoint condition and uses the model to reverse (from the receiver to the transmitter) the distribution of optical power in the optical fiber (as shown in Equation (9)).

[0123] The spectral tilt is the slope of the linear trend of the optical power spectrum with frequency in the decibel-milliwatt (dBm) or logarithmic power domain, typically measured in dBm / THz, and quantifies the degree of power transfer caused by the ISRS effect. The spectral tilt distribution refers to the curve showing the variation of the spectral tilt value along the fiber length direction (distance z).

[0124] In step 300, the forward evolution calculation will convert the transmit power spectrum { Substituting the (i=1, 2, ..., N) and link parameters into the closed-form solution of the power evolution model (as shown in Equation (6)), the signal power of each channel is calculated when the signal is transmitted from the transmitter along the optical fiber to any distance z, thus obtaining the forward power distribution. Subsequently, at each distance z, a linear regression is performed on the logarithm of all channels and the channel frequency, and the resulting regression slope is the forward spectral tilt at that location. The tilt values ​​of all distances z constitute a positive spectral tilt distribution.

[0125] Reverse evolution calculation is to calculate the received power spectrum obtained online { Substituting the calculated abnormal insertion loss values ​​and link parameters into the closed-form solution of the reverse power evolution model (as shown in Equation (9)), the estimated power of each channel at any distance z is calculated backward from the receiver to obtain the reverse power distribution. Similarly, linear regression is performed at each distance z to obtain the reverse spectrum tilt. This results in a reverse spectral tilt distribution.

[0126] In this application, power spectrum tilt is calculated using linear regression. However, alternative methods can be employed: weighted linear regression with different weights assigned to different channels; local spectral tilt calculation within a frequency band subset; or equivalent tilt indices based on the first or second moment of the power spectrum. These alternative methods essentially still measure the trend of power spectrum variation with frequency and do not deviate from the technical concept of this application's application example.

[0127] Furthermore, the forward power evolution and reverse power evolution described in the embodiments of this application are not limited to two independent calculation processes distinguished by the direction of optical signal propagation, but include power evolution models constructed based on different boundary conditions. As long as the observable power at both ends of the link is used as the constraint condition and the abnormal position is determined in the middle of the link through the consistency condition, they all fall within the protection scope of the application examples of this application.

[0128] The methods for calculating forward and reverse power include, but are not limited to, closed-form solutions, numerical methods, numerical simulations, and forward-reverse simulations. The consistency criteria for forward and reverse power evolution are not limited to strictly equal analytical conditions; they can also be based on minimizing error, minimizing distance, or maximizing correlation. These criteria are mathematically equivalent and are all used to determine the location of anomalies. The features used to determine the consistency of forward and reverse power evolution are not limited to power spectrum tilt, but also include statistical characteristics of the power spectrum, frequency domain characteristics, or equivalent metrics.

[0129] Step 400: Based on the characteristic that the forward spectral tilt distribution and the reverse spectral tilt distribution match the optical fiber anomaly location, determine the location of the anomaly in the target optical fiber link.

[0130] After step 400, the abnormal insertion loss value of the target optical fiber link obtained in step 200 and the location of the abnormality of the target optical fiber link obtained in step 400 can be output. For example, the abnormal insertion loss value and the location of the abnormality can be sent to a user-preferred client device for display, such as a computer, mobile phone, smartwatch, etc.

[0131] Specifically, the principle that spectral tilt remains constant at anomaly locations can be utilized to find a point where the forward and reverse spectral tilts are equal. This is achieved by substituting the closed-form expressions corresponding to the two spectral tilt distributions into the matching conditions, directly solving for the closed-form solution regarding the location point, and then substituting the known parameters to calculate the specific anomaly location coordinates. For example, if the calculated location point is 42.5 km, it indicates that the anomaly occurs approximately 42.5 kilometers from the transmitter of the target fiber optic link.

[0132] The characteristic of matching the forward and reverse spectral tilt distributions at the fiber optic anomaly location can be termed spectral tilt invariance. This means that while fiber optic anomalies cause power degradation, the signal power spectral tilt (dominated by the ISRS effect) before and after the anomaly occurs remains continuous without abrupt changes. This physical characteristic is the basis for localization. The matching can be mathematically defined as making the spectral tilt values ​​corresponding to the forward and reverse spectral tilt distributions equal to establish a solution equation.

[0133] As described above, the fiber optic anomaly detection method provided in this application can solve the problems of high cost and deployment flexibility caused by the introduction of additional dedicated hardware in traditional optical time-domain reflectometer schemes, thus eliminating the need for increased hardware overhead and facilitating easy integration and distributed deployment. By determining the anomaly loss value based on closed-loop calculation of the total power spectrum, it can solve the problem of dependence on massive anomaly training data for data-driven methods such as neural networks, thereby effectively reducing data requirements, avoiding model black-boxing, and improving applicability. By using a physical power evolution model that includes the stimulated Raman scattering effect between channels to calculate the forward and reverse spectral tilt distribution, it can solve the problems of computational complexity and large storage requirements of numerical inversion methods such as digital signal processing, making the calculation process simple, efficient, and physically interpretable. By locating based on the matching characteristics of the spectral tilt distribution at the anomaly location, it can solve the problem of existing methods being unable to balance detection real-time performance, positioning accuracy, and mechanism clarity, thereby achieving fast, accurate, and mechanism-transparent fiber optic anomaly location and diagnosis.

[0134] To further address the issue of decreased detection accuracy of fixed-parameter power evolution models applied to different real-world links due to individual differences in fiber optic links, device manufacturing process variations, and errors introduced by theoretical model simplification, this application provides a fiber optic anomaly detection method, see [link to relevant documentation]. Figure 4 The fiber optic anomaly detection method further includes the following steps prior to step 100:

[0135] Step 010: Optimize the key parameters in the power evolution model using training data generated from the digital twin model of the target fiber optic link, and configure the power evolution model with the optimized key parameters; wherein, the key parameters include: Raman gain slope, reference frequency, and equivalent attenuation index.

[0136] It should be noted that the target fiber optic link is a high-precision virtual replica in digital space. It is constructed based on the actual physical parameters of the link (such as length, attenuation coefficient, nonlinear coefficient, effective area and Raman gain spectrum shape, etc.), and can realistically simulate the transmission behavior of optical signals in the real link (as shown in Equation (1)) by solving optical transmission equations (such as the nonlinear Schrödinger equation), including attenuation, nonlinear effects (such as stimulated Raman scattering) and the effects of simulated abnormal events.

[0137] The training data is a dataset generated by a Digital Twin (DT) model to optimize key parameters. Each set of data simulates a specific anomalous scenario and typically includes: a simulated transmitter power spectrum, a simulated receiver power spectrum measured after a specific anomaly occurs, and the anomaly location and anomaly size labels corresponding to that set of data.

[0138] The key parameters refer to several parameters in the closed-form solution of the power evolution model that incorporates stimulated Raman scattering, which significantly affect the model's accuracy and may vary with the actual conditions of the link. Optimizing these parameters aims to make the simplified closed-form solution model approximate the behavior of the real link or a high-precision digital twin as closely as possible. Specifically, these include:

[0139] (1) Raman gain slope: The slope parameter used in the model when performing a triangular approximation of the stimulated Raman scattering gain spectrum (as shown in equation (4)) is used to characterize the rate at which the Raman gain between channels changes with the frequency difference.

[0140] (2) Reference frequency: A specific frequency value in the model used to characterize a reference point that is not affected by the spectral tilt caused by stimulated Raman scattering (as shown in Equations (3) and (6)). At this frequency, the channel power evolution is determined only by attenuation.

[0141] (3) Equivalent attenuation exponent: An adjustable positive parameter in the model, which works together with the fiber length to characterize the total power in the closed solution as an exponential attenuation along the fiber (as shown in equation (5)). Its optimization helps to compensate for the errors caused by model approximation.

[0142] Optimizing the key parameters in the power evolution model refers to automatically adjusting the values ​​of the key parameters using automated algorithms (such as particle swarm optimization) with the goal of minimizing the difference between the model's predicted output and the actual output generated by the digital twin, and finding the optimal parameter combination that best matches the current link. The closed-form solution of the power evolution model describes the analytical expression of the power distribution at any distance z in the optical fiber for a given emission spectrum (as shown in Equation (6)) and the corresponding reverse evolution form (as shown in Equation (9)).

[0143] Configuring the power evolution model involves fixing the optimal key parameter values ​​obtained from optimization and substituting them into all phase-closed formulas of the power evolution model (such as formulas (6), (9), etc.) to complete the personalized settings of the model, making it suitable for anomaly detection of the current specific fiber optic link.

[0144] As can be seen from the above description, the fiber optic anomaly detection method provided in this application, through digital twin-assisted parameter adaptive optimization, enables the core detection model to be automatically calibrated and adapted to any specific target link, significantly improving the accuracy, robustness and versatility of the method in real complex environments.

[0145] To further overcome the problems of high cost, poor real-time performance, lack of physical interpretability, and difficulty in distributed deployment caused by existing solutions relying on additional hardware, massive training data, or complex digital signal processing, this application provides a fiber optic anomaly detection method, see [link to relevant documentation]. Figure 5 Step 010 of the fiber optic anomaly detection method specifically includes the following:

[0146] Step 011: Construct the digital twin model based on the physical parameters of the target optical fiber link; wherein the physical parameters include at least one of the following: fiber length, attenuation coefficient, nonlinear coefficient, effective area, and Raman gain spectrum.

[0147] Specifically, the physical parameters of the actual fiber optic link to be tested are collected, and a high-fidelity simulation model, or digital twin model, is built in a computer. For example, for an 80km section of G.652.D fiber, the attenuation coefficients at different frequencies are input, given its length L=80km. Nonlinear coefficients Effective area and Raman gain spectrum By using parameters such as these, a system capable of simulating optical transmission can be constructed by solving precise numerical models such as the nonlinear Schrödinger equation (NLSE).

[0148] Step 012: Using the digital twin model, simulate multiple anomalous events to generate multiple sets of training data; wherein each anomalous event is defined by an anomalous location and an anomalous size, and each set of training data includes a corresponding simulated transmit power spectrum, simulated anomalous receive power spectrum, anomalous location label, and anomalous size label.

[0149] Specifically, in a digital twin model, a series of simulated abnormal events are systematically set up. For example, at intervals of 10km, in... At different anomaly locations (0, 10, 20, ..., 80 km), abnormal insertion loss values ​​(i.e., anomaly sizes) of β = 1, 2, 4, 6, 8, and 10 dB were introduced, respectively. Record the values ​​for each simulation run.

[0150] 1) Simulated transmit power spectrum, i.e., the simulated transmit power spectrum { , (i=1, 2, ..., N)} (i.e., input).

[0151] 2) Simulated abnormal received power spectrum { ,(i=1, 2, ..., N)} (i.e., the output after a simulation anomaly occurs).

[0152] 3) The location where the anomaly occurred in this simulation. The label for the abnormal size β.

[0153] This method generates multiple sets of training data that cover potentially abnormal scenarios.

[0154] Step 013: Based on the power evolution model that includes the inter-channel stimulated Raman scattering (ISRS) effect, determine the parameter set that includes the key parameters.

[0155] Specifically, the power evolution model is used to describe the evolution of optical power in multi-band optical fiber with propagation distance. Its closed-form solution for forward evolution calculation is shown by equation (6), and its closed-form solution for reverse evolution calculation is shown by equation (7).

[0156] As can be seen from equation (6), the accuracy of the power evolution model mainly depends on three key parameters: Raman gain slope. Reference frequency And decision The equivalent decay exponent n ( The calculation depends on n, as shown in equation (5). Therefore, the parameter set Θ to be optimized is determined as: Θ = [ , , n].

[0157] Step 014: With minimizing the loss function value as the optimization objective, iteratively optimize the parameter set to obtain the optimized parameter set; wherein, the loss function value characterizes the difference between the predicted power spectrum calculated by the power evolution model based on the simulated transmit power spectrum under the current value of the parameter set and the corresponding simulated abnormal receive power spectrum, as well as the difference between the spectral tilt of the predicted power spectrum and the spectral tilt of the corresponding simulated abnormal receive power spectrum.

[0158] Specifically, the Particle Swarm Optimization (PSO) algorithm is used to automatically search for the optimized parameter set, i.e., the optimal parameter set, with the goal of minimizing the model prediction error. Specifically, step 014 can employ a particle swarm optimization (PSO) algorithm to perform at least one iteration of optimization steps on the parameter set until a preset optimization termination condition is met, and the parameter values ​​in the parameter set after the termination iteration are taken as the optimized parameter set.

[0159] The optimization steps specifically include the following:

[0160] 1) Based on the initial parameter values ​​of the parameter set in the current iteration round, the power evolution model is used to calculate the simulated transmit power spectrum to obtain the predicted power spectrum for the current iteration round;

[0161] 2) Based on the predicted power spectrum of the current iteration round and the corresponding simulated abnormal received power spectrum, calculate the loss function value of the current iteration round; specifically, this can be: calculating a first error term and a second error term; wherein, the first error term represents the mean square error between the predicted power spectrum of the current iteration round and the corresponding simulated abnormal received power spectrum; and the second error term represents the error between the spectral tilt of the predicted power spectrum of the current iteration round and the spectral tilt of the corresponding simulated abnormal received power spectrum; determine the loss function value of the current iteration round based on the weighted sum of the first error term and the second error term.

[0162] 3) Based on the loss function value of the current iteration, update the parameter values ​​of the parameter set according to the update rules of the Particle Swarm Optimization (PSO) algorithm, so as to use them as the initial parameter values ​​in the next iteration.

[0163] In this application, a particle swarm optimization algorithm is used to optimize the parameter set in the closed-form solution. Without altering the core technical concept of this application, the parameter optimization process can also be implemented using one or a combination of the following alternatives: using a genetic algorithm, differential evolution algorithm, simulated annealing algorithm, or other intelligent optimization algorithm instead of the particle swarm optimization algorithm; using gradient descent, quasi-Newton method, or least squares-based parameter fitting methods to optimize the parameters; none of the above alternatives affect the closed-form solution detection principle itself and still fall within the protection scope of the application examples of this application.

[0164] Step 015: Configure the power evolution model based on the optimized parameter set.

[0165] Specifically, the optimized parameters are fixed and substituted as constants into all relevant closed-form formulas used in the entire detection scheme, including formulas (6) and (9) used for subsequent online detection. At this point, the power evolution model has completed calibration for the current target fiber optic link and can be used for online anomaly detection.

[0166] In one example, the execution process of step 014 is as follows:

[0167] 1) Initialization: Setting , The search range of n and PSO hyperparameters.

[0168] 2) Forward calculation: In each iteration, the value of the current parameter set Θ is substituted into formula (6). Using the simulated transmit power spectrum in a set of training data as input, the corresponding predicted power spectrum is calculated.

[0169] 3) Calculate the loss function value: Calculate the difference between the predicted power spectrum and the corresponding simulated anomalous received power spectrum in the training data. This difference value is quantified by the loss function. The loss function value is the first error term. With the second error term The weighted sum. The first error term. Characterized by the mean square error between the predicted power spectrum and the simulated anomalous received power spectrum; the second error term The error representing the spectral tilt between the two is used to characterize the difference. By adjusting the weighting coefficients, the impact of the two errors on the optimization process can be balanced. For details, please refer to equations (22) to (24) provided in the subsequent application examples.

[0170] 4) Iterative update: The PSO algorithm calculates the total loss function value based on all current training data, judges the quality of the parameters, and updates the particle positions (i.e., the parameter values ​​in the parameter set).

[0171] Repeat steps 1) to 4) above until the loss function value converges or the maximum number of iterations is reached. The parameter set obtained at this point is the optimized parameter set.

[0172] As can be seen from the above description, the optical fiber anomaly detection method provided in this application embodiment only utilizes existing optical performance monitoring data and achieves rapid and accurate anomaly location and quantification through a closed-form solution with a clear physical mechanism. It has the advantages of requiring no additional hardware, being computationally simple, having strong physical interpretability, good real-time performance, and being easy to deploy in a distributed manner.

[0173] To further address the problem of accurately separating and quantifying the additional insertion loss value introduced by fiber anomalies, independent of the inherent fiber loss, from the observed total power attenuation, an fiber anomaly detection method is provided in this application embodiment, see [link to relevant documentation]. Figure 4 Step 200 in the fiber optic anomaly detection method specifically includes the following:

[0174] Step 210: Calculate the first difference between the total power of the transmitted power spectrum and the total power of the received power spectrum; wherein the first difference is a value in decibels.

[0175] It should be noted that the first difference is a value expressed in decibels, representing the total apparent power attenuation experienced by the signal as it travels from the transmitter to the receiver. This attenuation includes both the inherent attenuation of the optical fiber and potential additional losses introduced by abnormal events.

[0176] In step 210, the power values ​​of all channels in the transmit power spectrum obtained in step 100 are summed to obtain the total transmit power. The power values ​​of all channels in the receive power spectrum obtained through online monitoring in step 100 are summed to obtain the total receive power. The total transmit power and the total receive power are converted to values ​​in decibel-milliwatts (dBm). The difference between these two decibel values ​​is calculated to obtain the first difference.

[0177] Step 220: Based on the optimized equivalent attenuation index, calculate the inherent attenuation of the target optical fiber link; wherein, the inherent attenuation is the power attenuation caused by the inherent characteristics of the optical fiber when the total power of the transmitted power spectrum is transmitted to the measurement position of the received power spectrum under the normal state of the target optical fiber link.

[0178] The inherent attenuation is a theoretically calculated value, representing the inevitable power loss of an optical fiber link under healthy conditions, determined by the fiber length and material properties. For example, an 80km long optical fiber with an equivalent attenuation factor of 0.2 dB / km has an inherent attenuation of approximately 16 dB.

[0179] Specifically, in step 220, the optimal equivalent attenuation index obtained through the aforementioned optimization process is acquired and substituted into formula (5). Combined with the known fiber attenuation coefficient and the continuous form of the transmit power spectrum, the optimized equivalent attenuation coefficient for the current link is calculated. The inherent attenuation is determined, which represents the power attenuation value caused solely by the inherent characteristics of the fiber material itself when the total transmit power is transmitted to the receiving end at a distance equal to the link length under ideal conditions where the target fiber link is in a fault-free state.

[0180] Step 230: Determine the abnormal insertion loss value based on the second difference between the first difference and the inherent attenuation.

[0181] The second difference is the abnormal insertion loss value, which refers to the additional power loss attributable to the abnormal event remaining after deducting the inherent attenuation of the link itself from the observed total attenuation.

[0182] Specifically, the execution of steps 210 to 230 above can be achieved through calculation formula (8).

[0183] As can be seen from the above description, the fiber optic anomaly detection method provided in this application embodiment can quickly and accurately distinguish and calculate the pure insertion loss caused by abnormal events, avoid misjudging the inherent attenuation of the link as an anomaly, and provide accurate anomaly size input for subsequent precise positioning.

[0184] To further address the issue of how to accurately calculate, using a physical model, the characteristic quantity (spectral tilt distribution) that characterizes the variation of the signal power spectrum shape (affected by ISRS) along the optical fiber, starting from the power spectra at both the transmitting and receiving ends, an optical fiber anomaly detection method is provided in this application embodiment, see [link to relevant documentation]. Figure 4 Step 300 in the fiber optic anomaly detection method specifically includes the following:

[0185] Step 310: Based on the power evolution model and the transmitted power spectrum, perform evolution calculations along the signal transmission direction to obtain the forward power distribution; wherein, the forward power distribution represents: the predicted signal power of each frequency channel at any position along the signal transmission direction in the target optical fiber link, starting from the transmitted power spectrum.

[0186] Specifically, the transmit power spectrum and optimized parameter set obtained in step 100 are acquired. Using the transmit power spectrum as initial conditions, the forward evolution closed-form solution of the power evolution model (i.e., equation (6)) is substituted to calculate the power of each channel at any propagation distance z as the signal propagates from the transmitter (z=0) along the optical fiber to the receiver. By changing the value of z (e.g., from 0 to the link length L, calculated in steps of a certain size), a series of... The set of values. This set is the forward power distribution, which fully characterizes the predicted signal power of each frequency channel at any position along the signal transmission direction in the target optical fiber link, starting from the transmit power spectrum.

[0187] Step 320: Perform linear regression on the logarithm of the signal power in the forward power distribution with frequency to calculate the forward spectrum tilt distribution; wherein, the forward spectrum tilt distribution characterizes: the regression slope of the logarithm of the signal power in the forward power distribution as a function of frequency at any position along the signal transmission direction.

[0188] Specifically, in order to calculate the spectral tilt during the power evolution process, we first take the logarithm ln of both sides of equations (6) and (7):

[0189] (10)

[0190] (11)

[0191] in, This represents the natural logarithm of the channel power at a propagation distance z; Represents the natural logarithm of the channel power in the transmit power spectrum; The total power attenuation factor is defined as the proportion of total signal power remaining when transmitted from the transmitter to a distance z, considering only the inherent attenuation described by the equivalent attenuation coefficient. It is represented as the natural logarithm of the inverse power estimate; It is represented as the natural logarithm of the received power spectrum obtained from online monitoring; The total power attenuation factor represents the proportion of total signal power remaining when transmitted from the transmitter to the end of the link (receiver), considering only the inherent attenuation described by the equivalent attenuation coefficient.

[0192] For the calculated set of channel powers at each specific propagation distance z, calculate the natural logarithm of each channel power. .Will Considered as the dependent variable, the i-th channel at its center frequency Treating the data as the independent variable, perform a linear regression on the data from all channels (i=1 to N). The slope of the regression line represents the positive spectral tilt at position z. Repeat this process for all the propagation distances z of interest, and obtain... The set that varies with z is the positive spectral tilt distribution.

[0193] in, The calculation formula is:

[0194] (12)

[0195] in, This represents the average value of all N channel frequencies.

[0196] And, step 330: based on the power evolution model and the received power spectrum, perform evolution calculations in the opposite direction to signal transmission to obtain the reverse power distribution; wherein, the reverse power distribution represents: the estimated signal power of each frequency channel at any position in the target optical fiber link along the signal transmission direction, starting from the received power spectrum.

[0197] Specifically, the received power spectrum obtained from online monitoring in step 100, the abnormal insertion loss values ​​β (and γ) calculated in step 200, and the optimized model parameters are obtained. Using the received power spectrum as the endpoint condition, the closed-form solution of the reverse power evolution applicable to abnormal scenarios (i.e., equation (9)) is substituted into it. Starting from the receiver (z=L), the power of each channel at any propagation distance z is calculated in reverse (i.e., along the direction where z decreases). Equation (9) is a generalization of equation (7) under abnormal scenarios. By changing the value of z, a set of channel power estimates is obtained. This set is the reverse power distribution, which characterizes the estimated signal power of each frequency channel at any position along the signal transmission direction in the target optical fiber link, starting from the (potentially abnormal) received power spectrum.

[0198] Step 340: Perform linear regression on the logarithm of the signal power in the reverse power distribution with frequency to calculate the reverse spectrum tilt distribution; wherein, the reverse spectrum tilt distribution characterizes: at any position along the signal transmission direction, the regression slope of the logarithm of the signal power in the reverse power distribution as a function of frequency.

[0199] For the calculated set of channel power estimates at each specific propagation distance z, calculate the natural logarithm of each channel power. .right With the i-th channel at its center frequency Performing linear regression yields the inverse spectral tilt at that distance z. Repeat this process for all the propagation distances z of interest, and you will get... The set that varies with z is the inverse spectral tilt distribution.

[0200] in, The calculation formula is as follows:

[0201] (13)

[0202] According to the principle of spectral tilt invariance, if and only if at the location of the anomaly At this point, the forward and reverse spectral tilts are equal.

[0203] As can be seen from the above description, the fiber optic anomaly detection method provided in this application provides a spectral tilt calculation method that is based on a closed-loop solution model, is computationally efficient, and has clear physical meaning. This provides the necessary key input features (i.e., forward and reverse spectral tilt distributions) for subsequent accurate anomaly localization based on spectral tilt invariance.

[0204] To further address the issues of high cost and difficult deployment of traditional hardware solutions (such as OTDR), and poor real-time performance, computational complexity, and lack of physical interpretation in data-driven or numerical inversion schemes, this application provides a fiber optic anomaly detection method, see [link to relevant documentation]. Figure 4 Step 400 in the fiber optic anomaly detection method specifically includes the following:

[0205] Step 410: Based on the forward spectral tilt distribution and the reverse spectral tilt distribution, calculate the location of the anomaly in the target optical fiber link using a closed-form solution constructed based on the matching condition of equal spectral tilt.

[0206] Specifically, based on the principle of spectral tilt invariance, the precise location of the fiber optic anomaly is determined. At this point, the spectral tilt obtained from forward evolution calculations should be equal to the spectral tilt obtained from backward evolution calculations. That is:

[0207]

[0208] This is the basic equation for determining the location of anomalies.

[0209] Substituting the formulas for calculating spectral tilt, namely equations (12) and (13), into the matching conditions above, and setting the right-hand side of the equation to be equal, we can obtain the location of the anomaly after substitution and simplification. The specific equation (i.e., the principle formula):

[0210] (14)

[0211] To solve equation (14), the following substitutions are performed to simplify the expression:

[0212] (15)

[0213] (16)

[0214] (17)

[0215] (18)

[0216] (19)

[0217] in, Represents a frequency-centered variable; Indicates the logarithmic power ratio; This represents the inherent attenuation term throughout the channel; Indicates the intensity factor of the ISRS effect; This represents the total power attenuation correction term related to anomalies.

[0218] Equation (14) can be rewritten as:

[0219] (20)

[0220] Solving equation (20) yields the closed-form solution for the location of the anomaly:

[0221] (twenty one)

[0222] Equations (8) and (21) are closed-form expressions for the size and location of optical fiber anomalies, respectively. As can be seen from the expressions, the proposed anomaly detection method can detect optical fiber anomalies by relying solely on the transmission and reception power. It does not depend on a large amount of training data or complex calculations, does not require a large amount of storage space to store signal waveform data, and is fast, fully physically interpretable, and can be distributed.

[0223] As can be seen from the above description, the fiber optic anomaly detection method provided in this application embodiment only utilizes existing optical performance monitoring data and achieves fast, accurate, and physically interpretable fiber optic anomaly localization and quantification without additional hardware or iterative search through a closed-form solution with a clear physical mechanism. It has the advantages of good real-time performance and suitability for distributed deployment.

[0224] To further illustrate the above embodiments, this application also provides a specific application example of an optical fiber anomaly detection method, specifically a derivation and implementation of a closed-form solution method for optical fiber link anomaly detection based on transmit and receive power spectra. This can be demonstrated in the following aspects:

[0225] 1. The application example of this application only utilizes the transmit and receive power of the signal transmitted across the fiber optic segment, without introducing additional hardware equipment and additional out-of-band probe optical pulses. It maintains the accuracy of fiber optic anomaly detection without interfering with or affecting the transmitted signal and without hardware deployment costs.

[0226] 2. The closed-form solution method in this application example can calculate the location and size of fiber optic anomalies using only the transmit and receive power of a pair of transmitted channels, thus solving the problem of neural network models requiring a large amount of data. Furthermore, this application example integrates the proposed closed-form solution method into a general fiber optic anomaly detection framework. This framework can be applied to any fiber optic link, enabling rapid migration and application to fiber optic links with different parameters, avoiding the generalization problem of neural network models.

[0227] 3. The application example of this application utilizes the characteristic that the spectral tilt remains unchanged when an optical fiber anomaly occurs, and derives a closed-form solution for detecting the location and size of an optical fiber anomaly using the closed-form solution of multi-band optical power evolution. It has obvious physical interpretability and fills the gap in the physical interpretability of optical link anomaly methods.

[0228] The anomaly detection scheme in this application example relies solely on the optical power value of the transmitted channel in the optical domain, eliminating the need for extensive storage of received signal data. Furthermore, the closed-loop computation method avoids the complex calculations inherent in DSP schemes, enabling the proposed scheme to achieve a fast response time once deployed. These characteristics also make the proposed scheme easier to deploy in a distributed manner within optical networks, thus resolving the real-time performance and deployment issues of both hardware and DSP detection schemes.

[0229] Because the derivation of equation (6) is based on the Raman gain spectrum triangular approximation and the assumption that the change in total power is determined only by a constant attenuation coefficient, there will be an error in the spectral tilt between the true solutions of equation (6) and equation (1). Furthermore, errors may also be introduced when using EDFA for lumped amplification and OCM acquisition at the end of the segment. This error can be corrected by adjusting the parameters in the closed-form solution. , and the free parameters in the equivalent attenuation coefficient n To reduce this. Therefore, this application example proposes a digital twin-assisted PSO parameter optimization method using link digital twin (DT) and particle swarm search (PSO) algorithms to adaptively and quickly optimize the parameter set Θ = [ , [, n] is used to adapt to parameter changes and accumulated errors in the current link, ensuring the effectiveness, accuracy, and robustness of the fiber optic link anomaly detection method. The specific process is as follows: Figure 6 As shown.

[0230] Figure 6 The figure shows the system implementation flowchart of the proposed closed-loop fiber optic anomaly detection method. First, it is necessary to construct an accurate digital twin of the current link through parameter refinement. The digital twin is used to simulate fiber optic anomalies and obtain transmit-receive power data pairs under different anomaly conditions. It should be noted that when simulating anomalies, the simulated anomaly location should cover as many fiber locations as possible, and the simulated anomaly size should cover as many possible anomaly sizes as possible, so as to ensure that PSO searches for the most suitable parameter set Θ for the current link. After collecting the fiber optic anomaly data pairs simulated by the digital twin, the PSO optimization stage begins. First, the optimization boundaries of the three parameters to be optimized in the parameter set are set, and the parameter set Θ and the hyperparameters of PSO optimization are initialized: particle number, particle position, sub-velocity, learning factor, inertia weight, etc. Then, the parameters in equation (6) are... , and n Set to the initial value, calculate the output power under the current parameters using equation (6). Calculate the loss function value using the current output power and the acquired output power, where the loss function is composed of a weighted average of the mean square error between the current output power and the true output power, and the mean square error of their spectral tilt, as defined in equations (22)-(24):

[0231] (twenty two)

[0232] (twenty three)

[0233] (twenty four)

[0234] in The first error term represents the mean square error between the current output power and the actual output power. However, due to the triangular approximation of the Raman gain spectrum, there is an error between the spectral tilt obtained from equation (6) and the spectral tilt of the actual solution. In this case, only the... The existence of multiple solutions may lead to a solution that is not applicable to the current optimal link. , and nThis affects the stability and robustness of the closed-form solution. Therefore, in the application example of this application, the mean square error of the spectral tilt between the current output power and the true output power is used as the second error term. This further restricts the optimization direction of the parameters, ensuring the stability and robustness of the closed-form solution calculation. The weights ω1 and ω2 are... and The weighted weights. This represents the spectral tilt of the predicted power spectrum calculated by the power evolution model under the current parameters based on the simulated transmit power spectrum, under the j-th set of training data in the optimization process. This represents the spectral tilt value calculated from the received power spectrum containing simulation anomalies, directly simulated by the digital twin model during the parameter optimization process, for the j-th set of training data. This represents the total number of training data sets used during parameter optimization. This indicates the total number of channels, that is, the total number of frequency channels contained in the transmit power spectrum and the receive power spectrum.

[0235] When the optimization algorithm fails to converge, the loss is calculated according to the loss function defined in equation (24) to determine the optimization of the parameters. In each iteration, the parameter set Θ and the optimization hyperparameters of PSO are updated, and the parameters are updated according to the parameter set Θ. , and n The loss is then recalculated until the algorithm converges or reaches the maximum number of iterations, thus obtaining the optimal parameters. c R , and n .

[0236] Optimal parameters , and n Substitute these values ​​into the closed-form solution for anomaly detection, equations (8) and (21). When an anomaly occurs in the actual link, the transmit power and abnormal receive power are collected by the OCM at both ends of the segment, and the closed-form solution can be used for fast and accurate detection.

[0237] This application example introduces simulated anomalies into an 80km optical transmission link and uses the proposed method and system to detect these anomalies. Specifically, fiber anomalies with a step size of 10km are introduced at distances from 0km to 80km, each with a step size of 2dB and a range from 1dB to 10dB. Optimal parameters are obtained through PSO optimization. , and n Then, the size of the anomaly is calculated using equation (8), and then the location of the anomaly is calculated together with equation (21). The detection results are shown in Figure 7(a) and Figure 7(b).

[0238] As shown in Figure 7(a), the calculation error of Equation (8) derived from the total channel power exhibits different calculation errors depending on the location and size of the anomaly, but all are within 0.1 dB. Meanwhile, Figure 7(b) shows the error diagram between the anomaly location calculated by Equation (21) and the actual anomaly location. It can be seen that when a large fiber optic anomaly occurs, a positioning error within 0.1 km can be achieved. For small anomalies, the error increases, but still remains within 0.5 km, demonstrating the effectiveness and robustness of the proposed method.

[0239] It should also be noted that the system architecture for implementing the method shown in the application examples of this application can have various deployment and implementation forms, including but not limited to: software modules integrated into optical network management systems; embedded algorithm modules deployed in coherent optical modules or circuit boards at the receiving end; and centralized or distributed anomaly detection services on cloud or edge computing nodes. The above-mentioned different system implementation methods differ only in deployment form and computing location, and do not change the essence of the technical solution proposed in the application examples of this application, and all fall within the protection scope of the application examples of this application.

[0240] In other words, this application example proposes a closed-loop anomaly detection method based solely on the power spectrum evolution caused by inter-channel stimulated Raman scattering (ISRS) in multi-band optical fibers. This method relies solely on the transmit and receive power spectra for the joint estimation of anomaly location and magnitude. A corresponding, engineering-feature-friendly anomaly detection system framework is also constructed. This method does not rely on additional hardware, requires no large-scale training data, and does not involve complex digital signal processing calculations. It possesses advantages such as good physical interpretability, real-time performance, and distributed deployment. Specifically, this application example provides the following improvements:

[0241] 1. Fiber optic anomaly detection principle based on power spectrum tilt invariance: This application example is the first to utilize the physical property that "when an optical fiber anomaly occurs, the power spectrum tilt remains unchanged before and after the anomaly", transforming the anomaly detection problem from a traditional signal time domain or complex numerical inversion problem into an analytical problem based on power spectrum distribution and tilt matching. This is the key theoretical basis for realizing closed-form solution derivation.

[0242] 2. Closed-form calculation method for fiber optic anomaly size: Based on the principle of total power conservation and attenuation superposition, this application proposes a closed-form expression that can calculate the anomaly insertion loss size using only the total transmitted power and the total received power of the anomaly, thus achieving fast and low-complexity estimation of the anomaly size.

[0243] 3. Closed-form solution calculation method for fiber optic anomaly location: In this application example, the power spectrum obtained by forward and reverse power evolution is linearly fitted in the logarithmic domain, and the analytical expression of the anomaly location is derived by using the constraint condition of equal spectral tilt. This realizes the closed-form solution of the anomaly location, rather than numerical search, neural network training or iterative inversion.

[0244] 4. Digital twin-assisted parameter adaptive optimization mechanism: To address the problem of model approximation error and system error accumulation in actual links, this application proposes a parameter adaptive optimization scheme based on link digital twin and particle swarm optimization algorithm, which is used to optimize the key parameter set in the closed solution online or offline, and improve the accuracy and robustness of the method under different link configurations.

[0245] 5. A fiber optic anomaly detection system framework suitable for distributed deployment without additional hardware: This application example constructs an anomaly detection framework that can run solely based on the transmit and receive power spectrum data collected by the existing optical performance monitoring equipment OCM. This avoids hardware detection schemes such as OTDR and highly complex processing schemes based on DSP, significantly reducing system cost and deployment difficulty, and improving real-time performance.

[0246] Based on this, compared with the prior art, the application examples of this application have the following beneficial effects:

[0247] 1. Regarding the amount of data training, although the method proposed in the application example of this application also requires data collection, only 10 or fewer data points are needed for the PSO algorithm to converge quickly and obtain the optimal parameters. At the same time, after the PSO search is completed, only the transmit and receive power spectra of the abnormal links are needed to realize the link anomaly detection. Therefore, it does not need to rely on a large amount of training data like data-driven methods, which greatly reduces the data dependency.

[0248] 2. The closed-form solution for the anomaly location derived in the application example of this application is based on the principle that the power tilt remains unchanged when there is an optical fiber anomaly. The location and size of the optical fiber anomaly are derived by using the closed-form solution of the forward and reverse evolution of the power of multi-band optical fibers. It has explicit physical interpretability. The application example of this application is the first to propose a closed-form solution for optical fiber anomaly detection, filling the gap in interpretable optical link anomaly detection methods.

[0249] 3. The derived closed-form solutions (8) and (21) for fiber optic anomaly detection only require the use of PSO to complete the optimal parameter search during the deployment phase. When performing fiber optic anomaly calculations, it only requires the use of the existing performance monitoring equipment OCM to collect the transmit and receive power spectrum, which takes about 200ms. Compared with the long response time of OTDR and the complex calculation based on DSP methods, the time cost of the proposed method is greatly reduced, ensuring the real-time performance of the proposed solution in the application example of this application.

[0250] 4. As shown in Equations (8) and (21), the closed-form solution method proposed in the application example of this application only requires the transmit and receive power pairs of the abnormal link and a limited number of additions and multiplications to calculate the location and size of the abnormality. It does not require a large amount of storage resources to store signal data and a large amount of computing power resources to perform complex DSP calculations and search algorithms, which greatly reduces the computing and storage resources consumed by anomaly detection and enables the proposed scheme to achieve distributed deployment.

[0251] 5. The application examples proposed in this application, such as Figure 6 The fiber optic link anomaly detection framework shown can adapt to fiber optic links with various configuration parameters. It can perform digital twin calibration on real links and use simulation data to achieve rapid optimal parameter search and transfer application, effectively avoiding the generalization problem of data-driven neural network methods.

[0252] From a software perspective, this application also provides an optical fiber anomaly detection device for performing all or part of the aforementioned optical fiber anomaly detection method, see [link to relevant documentation]. Figure 8 The fiber optic anomaly detection device specifically includes the following components:

[0253] The power acquisition module 10 is used to acquire the transmit power spectrum of the target optical fiber link under reference conditions, as well as the receive power spectrum obtained by online monitoring.

[0254] The anomaly size determination module 20 is used to determine the abnormal insertion loss value of the target optical fiber link based on the total power of the transmitted power spectrum and the total power of the received power spectrum;

[0255] The anomaly localization module 30 is used to perform forward evolution calculation on the transmitted power spectrum according to the power evolution model including the inter-channel stimulated Raman scattering (ISRS) effect, to obtain a forward spectrum tilt distribution, and to perform reverse evolution calculation on the received power spectrum, to obtain a reverse spectrum tilt distribution; based on the characteristic that the forward spectrum tilt distribution and the reverse spectrum tilt distribution match the anomaly location in the optical fiber, the anomaly location of the target optical fiber link is determined.

[0256] The embodiments of the optical fiber anomaly detection device provided in this application can be used to execute the processing flow of the embodiments of the optical fiber anomaly detection method described above. Its functions will not be repeated here, but can be referred to the detailed description of the embodiments of the optical fiber anomaly detection method described above.

[0257] The fiber optic anomaly detection device can perform the anomaly detection function in either a server or a client device. The choice depends on the processing power of the client device and the limitations of the user's usage scenario. This application does not impose any limitations in this regard. If all operations are performed in the client device, the client device may further include a processor for the specific processing of the fiber optic anomaly detection.

[0258] The aforementioned client device may have a communication module (i.e., a communication unit) that can communicate with a remote server to achieve data transmission with the server. The server may include a server on the task scheduling center side; in other implementation scenarios, it may also include a server on an intermediate platform, such as a server on a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, a server cluster consisting of multiple servers, or a distributed server structure.

[0259] The server and the client device can communicate using any suitable network protocol, including those not yet developed as of the date of this application. Such network protocols may include, for example, TCP / IP, UDP / IP, HTTP, HTTPS, etc. Furthermore, such network protocols may also include RPC (Remote Procedure Call Protocol) and REST (Representational State Transfer Protocol) protocols used on top of the aforementioned protocols.

[0260] As described above, the fiber optic anomaly detection device provided in this application only uses the transmit and receive power spectra at both ends of the target fiber optic link as input data. This solves the high cost and deployment flexibility issues associated with traditional optical time domain reflectometer (OTDR) schemes, which require additional dedicated hardware, thus eliminating the need for increased hardware overhead and facilitating easy integration and distributed deployment. By determining the anomaly loss value through closed-form calculation based on the total power of the power spectrum, it addresses the dependence of data-driven methods such as neural networks on massive amounts of anomaly training data, effectively reducing data requirements, avoiding model black-box architecture, and improving applicability. Furthermore, by employing a physical power evolution model incorporating inter-channel stimulated Raman scattering (ISRS) effects for forward and reverse spectral tilt distribution calculation, it solves the problems associated with digital signal processing (DSP). Numerical inversion methods such as Processing and Digital Spinning (DSP) are computationally complex and require large storage, making the computation process simple, efficient, and physically interpretable. By locating anomalies based on the matching characteristics of spectral tilt distribution at the location of anomalies, it can solve the problem that existing methods cannot simultaneously achieve real-time detection, positioning accuracy, and mechanistic clarity, thus enabling fast, accurate, and mechanistically transparent fiber optic anomaly localization and diagnosis.

[0261] This application also provides an electronic device, which may include a processor, a memory, a receiver, and a transmitter. The processor is used to execute the fiber optic anomaly detection method mentioned in the above embodiments. The processor and the memory can be connected via a bus or other means, taking a bus connection as an example. The receiver can be connected to the processor and the memory via wired or wireless means.

[0262] The processor can be a central processing unit (CPU). The processor can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or combinations of the above types of chips.

[0263] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules, such as the program instructions / modules corresponding to the fiber optic anomaly detection method in the embodiments of this application. The processor executes various functional applications and data processing by running the non-transitory software programs, instructions, and modules stored in the memory, thereby implementing the fiber optic anomaly detection method in the above method embodiments.

[0264] The memory may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created by the processor, etc. Furthermore, the memory may include high-speed random access memory and non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, the memory may optionally include memory remotely located relative to the processor, which can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0265] The one or more modules are stored in the memory, and when executed by the processor, the fiber optic anomaly detection method in the embodiment is executed.

[0266] In some embodiments of this application, the user equipment may include a processor, a memory, and a transceiver unit. The transceiver unit may include a receiver and a transmitter. The processor, memory, receiver, and transmitter may be connected via a bus system. The memory is used to store computer instructions, and the processor is used to execute the computer instructions stored in the memory to control the transceiver unit to send and receive signals.

[0267] As one implementation method, the functions of the receiver and transmitter in this application can be implemented by transceiver circuits or dedicated transceiver chips, and the processor can be implemented by dedicated processing chips, processing circuits or general-purpose chips.

[0268] As another implementation approach, the server provided in this application embodiment can be implemented using a general-purpose computer. That is, the program code implementing the processor, receiver, and transmitter functions is stored in memory, and the general-purpose processor implements the processor, receiver, and transmitter functions by executing the code in memory.

[0269] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the aforementioned fiber optic anomaly detection method. The computer-readable storage medium can be a tangible storage medium, such as random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, floppy disks, hard disks, removable storage disks, CD-ROMs, or any other form of storage medium known in the art.

[0270] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the aforementioned fiber optic anomaly detection method.

[0271] Those skilled in the art will understand that the exemplary components, systems, and methods described in conjunction with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Whether implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application. When implemented in hardware, it can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. The programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave.

[0272] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0273] In this application, features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, and / or combined with or in place of features of other embodiments.

[0274] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to the embodiments of this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for detecting optical fiber anomalies, characterized in that, include: Obtain the transmit power spectrum of the target fiber optic link under reference conditions, and the receive power spectrum obtained through online monitoring; Based on the total power of the transmitted power spectrum and the total power of the received power spectrum, the abnormal insertion loss value of the target optical fiber link is determined; Based on the power evolution model including the inter-channel stimulated Raman scattering (ISRS) effect and the transmitted power spectrum, evolution calculations are performed along the signal transmission direction to obtain the forward power distribution; wherein, the forward power distribution characterizes: the predicted signal power of each frequency channel at any position along the signal transmission direction in the target optical fiber link, starting from the transmitted power spectrum; A linear regression is performed on the logarithm of the signal power in the forward power distribution with frequency to calculate the forward spectral tilt distribution; wherein, the forward spectral tilt distribution characterizes: the regression slope of the logarithm of the signal power in the forward power distribution as a function of frequency at any position along the signal transmission direction; Furthermore, based on the power evolution model and the received power spectrum, an evolution calculation is performed in the opposite direction to the signal transmission to obtain a reverse power distribution; wherein, the reverse power distribution represents: the estimated signal power of each frequency channel at any position along the signal transmission direction in the target optical fiber link, starting from the received power spectrum; A linear regression is performed on the logarithm of the signal power in the reverse power distribution with frequency to calculate the reverse spectrum tilt distribution; wherein, the reverse spectrum tilt distribution characterizes the regression slope of the logarithm of the signal power in the reverse power distribution with frequency at any position along the signal transmission direction. Based on the forward spectral tilt distribution and the reverse spectral tilt distribution, the location of the anomaly in the target optical fiber link is calculated using a closed-form solution constructed based on the matching condition of equal spectral tilt. The closed-form solution for the location of the anomaly is: Indicates the location where the anomaly occurred; Represents a frequency-centered variable; Indicates the logarithmic power ratio; This represents the inherent attenuation term throughout the channel; Indicates the intensity factor of the ISRS effect; This represents the total power attenuation correction term related to anomalies; N is the number of channels, and i represents the i-th channel; This is the equivalent attenuation coefficient; Power ratio caused by anomalies in linear units The calculation formula is: ； This represents the total power of the received power spectrum; This represents the total power of the transmitted power spectrum; represents the abnormal insertion loss value in the logarithmic domain, i.e., the abnormal size; L represents the maximum length of the optical fiber.

2. The fiber optic anomaly detection method according to claim 1, characterized in that, Before obtaining the transmit power spectrum of the target fiber optic link under reference conditions, the method further includes: The key parameters in the power evolution model are optimized using training data generated from the digital twin model of the target fiber optic link, and the power evolution model is configured with the optimized key parameters; wherein, the key parameters include: Raman gain slope, reference frequency, and equivalent attenuation index.

3. The fiber optic anomaly detection method according to claim 2, characterized in that, The process of optimizing key parameters in the power evolution model using training data generated from the digital twin model of the target fiber optic link, and configuring the power evolution model with the optimized key parameters, includes: The digital twin model is constructed based on the physical parameters of the target fiber optic link; wherein, the physical parameters include at least one of fiber length, attenuation coefficient, nonlinear coefficient, effective area, and Raman gain spectrum; Using the digital twin model, multiple anomalous events are simulated to generate multiple sets of training data; wherein each anomalous event is defined by an anomalous location and an anomalous size, and each set of training data includes a corresponding simulated transmit power spectrum, simulated anomalous receive power spectrum, anomalous location label, and annomalous size label; Based on the power evolution model that includes the inter-channel stimulated Raman scattering (ISRS) effect, a parameter set containing the aforementioned key parameters is determined; The parameter set is iteratively optimized to obtain an optimized parameter set with the goal of minimizing the loss function value. The loss function value represents the difference between the predicted power spectrum calculated by the power evolution model based on the simulated transmit power spectrum under the current value of the parameter set and the corresponding simulated abnormal receive power spectrum, as well as the difference between the spectral tilt of the predicted power spectrum and the spectral tilt of the corresponding simulated abnormal receive power spectrum. Configure the power evolution model based on the optimized parameter set.

4. The fiber optic anomaly detection method according to claim 3, characterized in that, Determining the abnormal insertion loss value of the target optical fiber link based on the total power of the transmitted power spectrum and the total power of the received power spectrum includes: Calculate a first difference between the total power of the transmitted power spectrum and the total power of the received power spectrum; wherein the first difference is a value in decibels; Based on the optimized equivalent attenuation index, the inherent attenuation of the target optical fiber link is calculated; wherein, the inherent attenuation is the power attenuation value caused by the inherent characteristics of the optical fiber when the total power of the transmitted power spectrum is transmitted to the measurement position of the received power spectrum under the normal state of the target optical fiber link. The abnormal insertion loss value is determined based on the second difference between the first difference and the inherent attenuation.

5. The fiber optic anomaly detection method according to claim 3, characterized in that, The step of iteratively optimizing the parameter set to obtain an optimized parameter set by minimizing the loss function value includes: The Particle Swarm Optimization (PSO) algorithm is used to perform at least one iteration of optimization steps on the parameter set until the preset optimization termination condition is met, and the parameter values ​​in the parameter set after the termination iteration are used as the optimized parameter set. The optimization steps include: Based on the initial parameter values ​​of the parameter set in the current iteration round, the power evolution model is used to calculate the simulated transmit power spectrum to obtain the predicted power spectrum for the current iteration round. Based on the predicted power spectrum of the current iteration round and the corresponding simulated abnormal received power spectrum, calculate the loss function value of the current iteration round; Based on the loss function value of the current iteration, the parameter values ​​of the parameter set are updated according to the update rules of the Particle Swarm Optimization (PSO) algorithm, so as to serve as the initial parameter values ​​in the next iteration.

6. The fiber optic anomaly detection method according to claim 5, characterized in that, The calculation of the loss function value for the current iteration based on the predicted power spectrum of the current iteration and the corresponding simulated abnormal received power spectrum includes: Calculate a first error term and a second error term; wherein the first error term represents the mean square error between the predicted power spectrum of the current iteration round and the corresponding simulated abnormal received power spectrum; and the second error term represents the error between the spectral tilt of the predicted power spectrum of the current iteration round and the spectral tilt of the corresponding simulated abnormal received power spectrum. The loss function value for the current iteration is determined by the weighted sum of the first error term and the second error term.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the fiber optic anomaly detection method as described in any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the fiber optic anomaly detection method as described in any one of claims 1 to 6.

Images