Optical fiber end-to-end coverage fault location system based on passive optical switch cascade

Abstract

The application discloses a passive optical switch cascade-based optical fiber end-to-end coverage fault positioning system and relates to the technical field of optical fiber communication fault detection.The system comprises a mapping construction module, a first identification module, a second identification module and a warning module.The mapping construction module is used for establishing a mapping matrix of switch states and optical fiber sections based on a preset switch switching sequence.The first identification module is used for injecting a probe pulse under different switch states and inversing local time delay distortion of each section by using a time delay observation set to form a first type of structure representation.The second identification module is used for collecting an amplitude and phase sequence of an end-to-end signal, calculating a statistical distance of the amplitude and phase sequence from a reference and projecting the statistical distance to a section level space to form a second type of structure representation.The warning module is used for performing damage analysis under multi-source observation and reporting a fault anomaly.The technical problem that the prior art cannot realize synchronous fine fault perception and positioning of multiple potential deterioration sections in the whole end-to-end process is solved, and the technical effect that the health condition of an optical fiber link is perceived in real time by using a passive cascade switch network and the fault positioning precision is improved is achieved.

Description

Technical Field

[0001] This invention relates to the field of optical fiber communication fault detection technology, specifically to an end-to-end optical fiber coverage fault location system based on passive optical switch cascading. Background Technology

[0002] Fiber optic links typically span long distances and complex geographical environments. Physical layer damage such as bending, compression, aging, connector deterioration, or human-caused damage can easily lead to signal attenuation, time delay distortion, and even communication interruption. Traditional fiber optic fault detection is based on optical time domain reflectometers, whose measurement accuracy is limited by spatial resolution. In complex optical networks with long distances, multiple nodes, and dynamic changes, it is difficult to achieve refined, real-time, and low-cost end-to-end coverage monitoring, especially in distinguishing closely adjacent fault points or simultaneously monitoring multiple potentially deteriorated sections. In addition, existing fiber optic fault detection methods mostly rely on active monitoring equipment or distributed fiber optic sensing, which requires the insertion of electro-optic devices in the link, introducing additional fault points, power consumption, and cost. This results in high system complexity, expensive demodulation equipment, and insensitivity to faults outside the link terminal, making it impossible to simultaneously identify and locate multiple potentially faulty sections.

[0003] Therefore, current technologies face the technical challenge of simultaneously and precisely detecting and locating faults in multiple potentially deteriorated sections throughout the entire end-to-end process. Summary of the Invention

[0004] This application provides an end-to-end fiber optic coverage fault location system based on passive optical switch cascading, which solves the technical problem in the prior art that it is impossible to synchronously and accurately detect and locate multiple potentially degraded sections throughout the entire end-to-end process. It achieves the technical effect of real-time perception of the health status of the fiber optic link and improved fault location accuracy through a passive cascaded switch network.

[0005] This application provides an end-to-end fiber optic coverage fault location system based on cascaded passive optical switches. The system includes: a mapping construction module, used to deploy multiple cascaded passive optical switches in the fiber optic link to be tested, switch the passive optical switches at the endpoint control end according to a preset perturbation sequence, and establish a mapping matrix between the switch states and the fiber segments; a first identification module, used to inject time-coded probe pulses from the endpoint in each switch state to obtain end-to-end round-trip delay observation sets for different switch states, construct a segment-level delay inversion model based on the mapping matrix, solve for the local delay distortion of each fiber segment, and construct a first type of structured representation; a second identification module, used to synchronously acquire the amplitude and phase sequences of the end-to-end received signals in different switch states, calculate the statistical distance between the synchronous acquisition results and the reference sequence to form a switch-state statistical distance sequence, project the switch-state statistical distance sequence onto the segment-level space according to the mapping matrix, and construct a second type of structured representation; and an early warning module, used to perform damage analysis under multi-source observation based on the first type of structured representation and the second type of structured representation, and report fault anomalies.

[0006] In a possible implementation, the fiber optic end-to-end coverage fault location system based on passive optical switch cascading further includes: a third identification module, used to perform complex field detection on the end-to-end received signal in each switching state, construct a complex envelope sequence, construct a complex field cross-coherence function for any two switching states based on the complex envelope sequence, perform time-frequency analysis, extract coherent texture maps and calculate texture degradation indices, and construct a third type of structured representation after projection based on the mapping matrix; and a location verification module, used to perform fault trust authentication based on the third type of structured representation, reconstruct the fault anomaly level based on the fault trust authentication result, and report the fault location.

[0007] In a possible implementation, the third identification module is further configured to: perform gated coherent differential processing based on the complex envelope sequence; in each switching state, using the complex field phase of the reference state as the coherent anchor point, perform joint differential processing in the time and frequency domains through a gated window to obtain a differential coherent field that suppresses the noise of the entire link common factor; input the differential coherent field to a multi-scale complex wavelet packet decomposer, perform amplitude-phase coupled feature decomposition on the complex field perturbations of different switching states at multiple scales, and extract scale feature clusters that reflect the perturbation response of scatterers within the fiber segment; construct an intra-segment perturbation fingerprint matrix based on the mapping matrix between the scale feature clusters and the switching states, and perform sparse inversion based on spectral domain reversible reconstruction to obtain the perturbation-dominant scale and effect weights within each fiber segment, generating deep coherent texture features; and output the deep coherent texture features as an enhancement term of the third type of structured representation.

[0008] In a possible implementation, the first identification module constructs a segment-level time delay inversion model by: constructing a time delay observation vector based on the round-trip time delay observation set, and generating a segment-level observation operator corresponding to the switching state and the fiber segment based on the mapping matrix; introducing a pulse diffusion kernel characterization factor into the segment-level observation operator to construct an extended mapping matrix that can simultaneously describe the segment-level time delay accumulation effect and the pulse waveform diffusion effect; using the segment-level time delay distortion as the solution objective, the extended mapping matrix as the forward operator, and the time delay observation vector as the observation operator input to construct a segment-level time delay inversion model.

[0009] In a possible implementation, the second identification module is used to: perform amplitude-phase joint unification processing on the amplitude sequence and phase sequence of the end-to-end received signal, construct an amplitude-phase coupled vector sequence through phase expansion, amplitude normalization and group delay compensation; calculate the distribution divergence distance, phase deviation distance and complex domain energy distribution distance at multiple scales based on the amplitude-phase coupled vector sequence and the reference sequence, and combine the multi-scale distances to construct a switch-state statistical distance sequence.

[0010] In a possible implementation, the early warning module reports fault anomalies by: if the damage confidence exceeds a preset threshold, sending a feedback command to the endpoint control terminal to trigger a local enhanced scan of the switch disturbance sequence. The local enhanced scan includes dynamically combining switch states in the fault anomaly segment to generate a highly focused sub-mapping matrix, and performing secondary attention acquisition analysis based on the sub-mapping matrix to establish an enhanced observation set; after performing anomaly verification of the fault anomaly segment based on the enhanced observation set, reporting the fault anomaly.

[0011] In a possible implementation, the early warning module is further configured to: construct an end-to-end residual spectrum based on historical observation sets of different time windows; perform adaptive noise floor reconstruction based on the spectral density drift of the end-to-end residual spectrum; adjust the joint analysis weights of the first type of structured characterization and the second type of structured characterization according to the reconstructed noise floor to suppress the influence of noise floor changes on segment-level time delay inversion stability and amplitude-phase statistical distance; and perform damage analysis under multi-source observations according to the joint analysis weights.

[0012] In a possible implementation, the fiber optic end-to-end coverage fault location system based on cascaded passive optical switches further includes: a fault isolation module, used to reconfigure the on / off state of the passive optical switches according to the fault anomaly, and to perform fault isolation management.

[0013] This application proposes a fiber optic end-to-end coverage fault location system based on cascaded passive optical switches. The system comprises a mapping construction module that establishes a mapping matrix between switch states and fiber segments based on a preset switch switching sequence; a first identification module that injects probe pulses under different switch states and uses a time delay observation set to invert the local time delay distortion of each segment, forming a first type of structural representation; a second identification module that collects the amplitude and phase sequences of the end-to-end signal, calculates its statistical distance to a reference, and projects it onto the segment-level space, forming a second type of structural representation; and an early warning module that performs damage analysis under multi-source observation and reports fault anomalies. This system solves the technical problem in existing technologies where multiple potentially degraded segments throughout the entire end-to-end network cannot be simultaneously and precisely detected and located for faults. It achieves the technical effect of real-time perception of the health status of fiber optic links and improved fault location accuracy through a passive cascaded switch network. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments of this disclosure will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0015] Figure 1 This is a schematic diagram of the fiber optic end-to-end coverage fault location system based on cascaded passive optical switches, provided as an embodiment of this application.

[0016] Figure 2 This is a schematic diagram illustrating the execution steps of the third identification module in the fiber optic end-to-end coverage fault location system based on cascaded passive optical switches provided in this application embodiment.

[0017] Explanation of reference numerals in the attached diagram: Mapping construction module 10, first identification module 20, second identification module 30, early warning module 40. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by the present invention to achieve the intended purpose, the following detailed description is provided in conjunction with the accompanying drawings and preferred embodiments, based on the specific implementation methods, structures, features, and effects of the present invention. The technical solution of the present invention involves cascading passive optical switches in an optical fiber link, controlling the switch state through the endpoints, and establishing a mapping matrix between the switch state and the optical fiber segment. Subsequently, by measuring the round-trip delay and amplitude / phase changes under different switch states, structured representations of the first type of delay distortion and the second type of amplitude / phase statistics are constructed respectively. Finally, these are fused together for damage analysis and fault location and early warning.

[0019] This application provides an embodiment of an end-to-end fiber optic coverage fault location system based on cascaded passive optical switches, such as... Figure 1 As shown, the system includes: The mapping construction module 10 is used to deploy multiple cascaded passive optical switches in the optical fiber link to be tested, and switch the passive optical switches at the endpoint control end according to a preset perturbation sequence to establish a mapping matrix between the switch state and the optical fiber segment.

[0020] Preferably, the mapping construction module deploys passive optical switch hardware, executes orderly state switching control, and ultimately establishes a mathematical mapping matrix describing the state-path relationship, so as to perform segment-level fault inversion based on end-to-end measurements. Specifically, in the optical fiber link to be tested, multiple passive optical switches, such as MEMS optical switches or thermo-optical switches, are physically connected in a cascade manner. Each optical switch has two or more ports, and by controlling its state, different paths for the optical signal can be selected. The cascaded passive optical switches are arranged in key nodes such as junction boxes and distribution points of the optical fiber link, thereby dividing the entire optical fiber link into multiple independent optical fiber segments.

[0021] Preferably, a control terminal, such as a signal injection terminal, is set at one end of the optical fiber link. This control terminal sends optical signals as control signals to each passive optical switch according to a preset perturbation sequence with specific coding rules, so as to systematically switch the physical state of each switch, such as "straight-through" or "cross-through". The preset perturbation sequence ensures that the combination of different switch states can uniquely change the combination of optical fiber segments through which the optical signal passes. Then, based on the physical connection relationship of the passive optical switches and the preset perturbation sequence, a mapping matrix M between the switch state and the optical fiber segment is established. The rows of the matrix correspond to different switch state combinations, and the columns of the matrix correspond to each independent optical fiber segment. The matrix element M(i,j) takes the value of 1 or 0, which is used to define whether the optical signal passes through the j-th optical fiber segment in the i-th switch state, thereby completely and accurately describing the mapping relationship between the switch state combination and the set of optical fiber segments through which the optical signal passes.

[0022] The first identification module 20 is used to inject time-coded probe pulses from the endpoint in each switching state to obtain end-to-end round-trip delay observation sets for different switching states, construct a segment-level delay inversion model based on the mapping matrix, solve the local delay distortion of each fiber segment, and construct the first type of structured representation.

[0023] Preferably, the first identification module measures the macroscopic time delay under controllable switching states and uses an inversion model established by mapping relationships to decompose and attribute the macroscopic observations to each microscopic fiber segment, thereby quantitatively constructing a time delay anomaly distribution in segments. Specifically, under each preset switching state, one or more time-coded probe light pulses are injected into the fiber link from the end of the fiber link. The pulse signal propagates in the link and returns to the injection end after being reflected or looped back at the end. The end-to-end round-trip time delay or propagation time of the signal from injection to return is accurately measured and recorded under each switching state. After measuring and recording all preset switching states, an end-to-end round-trip time delay observation set for different switching states is obtained, which corresponds one-to-one with the switching state.

[0024] Preferably, a segment-level time delay inversion model is constructed based on a mapping matrix. The essence of the segment-level time delay inversion model is to solve a set of equations: Observation delay vector = mapping matrix × segment-level time delay vector + noise / error term. Each element of the observation delay vector corresponds to an end-to-end round-trip time delay observation value in a switching state. The mapping matrix is ​​used to describe the fiber segment through which the optical signal passes in each switching state. Each element of the segment-level time delay vector is an unknown quantity to be solved, representing the local time delay contribution or time delay distortion of each independent fiber segment. The segment-level time delay inversion model is solved using the least squares method or sparse recovery algorithm. From the macroscopic end-to-end time delay observation data, the local time delay distortion of each fiber segment is calculated. The finally calculated segment-level time delay distortion constitutes the first type of structured representation of the link state. Each element corresponds to a specific fiber segment, and its value reflects the signal propagation time anomaly caused by physical stress, temperature change, microbending, or damage in that fiber segment.

[0025] The second identification module 30 is used to synchronously acquire the amplitude sequence and phase sequence of the end-to-end received signal under different switching states, calculate the statistical distance between the synchronous acquisition result and the reference sequence, form a switching state statistical distance sequence, and project the switching state statistical distance sequence to the segment-level space according to the mapping matrix to construct the second type of structured representation.

[0026] Preferably, the second identification module analyzes the statistical deviation of the amplitude and phase of the signal relative to the reference under multiple switching states, and uses a projection model established by mapping relationship to trace the macroscopic and mixed deviations to each microscopic fiber segment, thereby quantitatively constructing the amplitude and phase anomaly distribution on a segment-by-segment basis. Specifically, under each preset specific switching state, the optical signal transmitted through the link is synchronously collected from the receiving end of the optical fiber link, and then the received optical signal is processed to extract its amplitude sequence and phase sequence that change with time or frequency, respectively characterizing the signal intensity change and signal phase change, and thus jointly reflecting the comprehensive effects such as loss, dispersion, and nonlinearity experienced by the signal during propagation.

[0027] Preferably, for each switching state, the amplitude sequence and phase sequence of the end-to-end received signal are compared with a predefined reference sequence, where the reference sequence represents the reference signal characteristics of the link under known healthy or undisturbed conditions. Then, the statistical distance between the synchronous acquisition result and the reference sequence is calculated, such as calculating the correlation coefficient, root mean square error, Kullback-Leibler divergence, or Euclidean distance in the complex domain between the two sets of sequences. By calculating, a corresponding quantized statistical distance is generated for each switching state, which characterizes the degree of deviation of the link signal characteristics under the current switching state from the reference sequence state. The statistical distance calculation is performed on all switching states to obtain a switching state statistical distance sequence that corresponds one-to-one with the switching state sequence.

[0028] Preferably, the statistical distance sequence of switch states defined in the switch state space is transformed into the fiber segment space through mathematical inversion projection based on the mapping matrix. Specifically, the statistical distance of each switch state is the result of the superposition of the combined effects of multiple fiber segments traversed by the optical path corresponding to that switch state. Then, using the switch state-fiber segment mapping relationship described by the mapping matrix, a constrained linear model is constructed to decompose the macroscopic switch state distance sequence, solve for and determine the contribution weight or local offset of each independent fiber segment to the statistical distance, and finally construct the solution results into a second type of structured representation. Each element of this representation corresponds to a specific fiber segment, and its value is used to quantify the contribution of that fiber segment to the overall signal amplitude and phase distortion, thereby reflecting the loss anomaly, phase disturbance, and other state information of that fiber segment. Among them, the second type of structured representation of amplitude and phase statistics and the first type of representation of time delay distortion are complementary in the physical dimension and jointly describe the health status of the fiber link.

[0029] The early warning module 40 is used to perform damage analysis under multi-source observation based on the first type of structured characterization and the second type of structured characterization, and to report fault anomalies.

[0030] Preferably, the early warning module integrates segment-level inversion results from two independent observation dimensions, time delay and amplitude phase, and performs cross-validation and comprehensive analysis to achieve fault location and alarm with higher reliability and lower false alarm rate. Specifically, it receives the first type of structured representation characterizing the local time delay distortion vector of each fiber segment and the second type of structured feature characterizing the amplitude phase statistical anomaly contribution vector of each fiber segment. It performs time synchronization and alignment on the segment-level data of time-domain propagation characteristics and amplitude phase transmission characteristics to ensure that they correspond to the same monitoring period and link status. Subsequently, it performs damage analysis under multi-source observation, including correlation analysis, contradiction identification, pattern recognition and confidence assessment.

[0031] Preferably, correlation analysis is used to determine whether the same fiber segment exhibits abnormal indicators simultaneously in the first and second types of characterization. For example, if a fiber segment simultaneously shows a significant increase in time delay and a sharp increase in amplitude statistical distance, then the segment has physical damage, such as severe bending or compression. Inconsistency identification is used to analyze whether there are physically inconsistent indications between the two types of characterization. For example, if a segment has a large time delay distortion but a small change in amplitude statistical distance, it may indicate that the abnormality is caused by a uniform change in refractive index, such as a temperature gradient. Pattern recognition refers to combining the characterization values ​​of each fiber segment into a multi-dimensional feature vector, and using a classification model pre-trained based on support vector machines or neural networks or a decision tree based on physical rules to identify whether its feature pattern belongs to "normal", "slight degradation", "suspected fault" or "definite fault", while outputting a damage confidence score for each fiber segment's fault anomaly determination. When the damage confidence of any fiber segment exceeds a preset threshold, a fault alarm signal is generated and sent. This includes identifying the specific fiber segment number or location that is determined to be abnormal or faulty, and reporting the type and severity level of the fault based on pattern recognition results and damage confidence scores. This ensures real-time awareness of the health status of the fiber link and improves the accuracy of fault location.

[0032] Furthermore, the specific configuration of the fiber optic end-to-end coverage fault location system based on passive optical switch cascading also includes: a third identification module, used to perform complex field detection on the end-to-end received signal in each switching state, construct a complex envelope sequence, construct a complex field cross-coherence function for any two switching states based on the complex envelope sequence, perform time-frequency analysis, extract coherent texture maps and calculate texture degradation indices, and construct a third type of structured representation after projection based on the mapping matrix; and a location verification module, used to perform fault trust authentication based on the third type of structured representation, reconstruct the fault anomaly level based on the fault trust authentication result, and report the fault location.

[0033] Preferably, the third identification module provides third-dimensional diagnostic information that is highly sensitive to microscopic consistency disturbances in the fiber optic link by analyzing the coherence texture between signals from different optical paths. Specifically, in each preset switching state, complex field detection is performed on the optical signals received end-to-end, including measuring the signal intensity and phase, thereby obtaining a complex form signal containing real and imaginary part information. This complex form signal is then processed by Hilbert transform to construct a complex envelope sequence describing the evolution of its amplitude and phase with events / frequency, so as to completely preserve the amplitude and phase information of the optical signal. Then, using the complex envelope sequences obtained in different switching states, a complex field cross-coherence function is calculated for signals in any two switching states. The complex field cross-coherence function is used to quantify the amplitude and phase correlation between the received signals in two different optical paths defined by different switching states.

[0034] Preferably, short-time Fourier transform or wavelet transform is used to perform time-frequency analysis on the complex field cross-coherence function, transforming the one-dimensional complex field cross-coherence function into a two-dimensional time-frequency joint distribution map, i.e., a coherent texture map, to reveal how coherence changes with time and frequency. Then, quantitative indicators that can characterize its structural properties are extracted from the coherent texture map to determine texture degradation indicators, such as the width and symmetry of the coherence peak, the energy distribution entropy value of the time-frequency plane, and the roughness of the texture. These indicators are extremely sensitive to microscopic perturbations such as Rayleigh scatterer changes, nonlinear effects, and polarization state perturbations in the fiber link. Finally, using the mapping relationship between the switch state and the fiber segment described by the mapping matrix, the texture degradation indicator sequence calculated based on the switch state pair is transformed into the fiber segment space through mathematical inversion projection, and the contribution of each independent fiber segment to the overall coherent texture degradation characteristics is determined as the third type of structured characterization.

[0035] Preferably, the location verification module independently verifies the initial warning using the third type of structured representation. Through fusion decision-making, it improves the accuracy of fault diagnosis, reduces false alarms, and outputs a more reliable and refined final fault location result. Specifically, it receives the third type of structured representation and cross-compares and verifies the anomalies of the third type of representation in the corresponding suspicious fiber segment with the anomalies of the first and second types of structured representation. If the third type of structured representation also shows a clear anomaly at the same location, and is physically explainable by the first and second types of anomalies (e.g., increased latency, increased loss, and a sharp decrease in coherence), then the authentication is successful, increasing the confidence in the existence of a real physical fault at that location. If the third type of structured representation shows normal or only a weak anomaly at the suspicious location, then a more in-depth analysis is triggered or the fault confidence at that location is reduced. Based on the trust authentication result, the fault anomaly level is updated and reconstructed; for example, a fault jointly confirmed by the three representations is upgraded to a high-confidence fault. Finally, a verified fault location report is output, including the fault location, the updated fault level, a consistency description of multi-source evidence, and detailed information about the nature of the fault provided by coherent texture analysis.

[0036] Furthermore, such as Figure 2As shown, the specific configuration of the third identification module further includes: performing gated coherent differential processing based on the complex envelope sequence; using the complex field phase of the reference state as the coherent anchor point in each switching state; performing joint differential processing in the time and frequency domains through a gated window to obtain a differential coherent field that suppresses the noise of the entire link common factor; inputting the differential coherent field to a multi-scale complex wavelet packet decomposer; performing amplitude-phase coupled feature decomposition on the complex field perturbations of different switching states at multiple scales to extract scale feature clusters that reflect the perturbation response of scatterers within the fiber segment; constructing an intra-segment perturbation fingerprint matrix based on the mapping matrix between the scale feature clusters and the switching states; and performing sparse inversion based on spectral domain reversible reconstruction to obtain the perturbation-dominant scale and effect weights within each fiber segment, generating deep coherent texture features; and outputting the deep coherent texture features as an enhancement term for the third type of structured representation.

[0037] Preferably, gated coherent differential processing is performed with the complex envelope sequence of each switching state as input. One switching state is selected as the reference state, and its complex field phase sequence is used as a stable phase reference, i.e., the "coherent anchor point". For each other switching state, its complex envelope sequence is compared with the reference state sequence. Joint differential processing is performed in the time domain and frequency domain through a gated window. That is, within the gated window, the complex field phase difference between the current switching state and the reference state is calculated, including the amplitude difference and the phase difference. The differential operation suppresses the noise and slow-varying disturbances common to the entire fiber link, such as laser phase noise and overall length changes caused by ambient temperature, as the common factor noise of the entire link. The differential coherent field is output, highlighting the signal changes corresponding to the optical path differences introduced by the switching state switching.

[0038] Preferably, the differential coherent field is input to a multi-scale complex wavelet packet decomposer to perform fine analysis of the time-frequency structure of the signal at different frequency bandwidths and time windows. Amplitude-phase coupling feature decomposition is performed on the complex field perturbations of different switching states at multiple scales. That is, the multi-scale complex wavelet packet decomposer simultaneously analyzes the coupling relationship between amplitude and phase changes in the differential coherent field at each scale. For example, the perturbations of scatterers in the optical fiber will affect the amplitude and phase of the signal in a specific way. Then, from the decomposition results of all scales, a set of features that can reflect the response characteristics of scatterers or structural perturbations inside the optical fiber segment is extracted as a scale feature cluster to characterize the amplitude-phase coupling perturbation mode of different switching state changes distributed at multiple scales. Each scale feature may correspond to statistics such as energy and correlation at different scales.

[0039] Preferably, a correlation model between scale feature clusters and each fiber segment is established using the mapping matrix of each switching state. This model represents the extracted multi-scale features, which are the perturbations within each fiber segment. These perturbations are linearly or nonlinearly superimposed according to their on / off states under different switching states. This results in the construction of an intra-segment perturbation fingerprint matrix, where rows represent fiber segments, columns represent corresponding scale features, and matrix elements represent the "weights" of the perturbations generated by that fiber segment at that feature scale. Then, based on spectral domain reversible reconstruction, sparse inversion is performed to solve for the perturbation fingerprint within each fiber segment, obtaining the dominant perturbation scale and effect weights within each fiber segment, i.e., assuming true... The actual faults or disturbances occur only in a few fiber segments, and within these segments, only a few characteristic scales are dominant. The solutions are sparsity, which allows for the estimation of the dominant disturbance scales and their corresponding effect weights within each fiber segment. This leads to the output of deep coherent texture features, which are used to indicate the main time-frequency scale characteristics and intensity of disturbances within anomalous fiber segments and their internal disturbances. Finally, the deep coherent texture features are output as an enhancement term for the third type of structured characterization, thereby greatly enhancing the resolution, anti-interference capability, and diagnostic depth of the third type of characterization, enabling it to detect and characterize weaker and earlier distributed damage or local defects.

[0040] Furthermore, the specific configuration of the first identification module 20 also includes: constructing a time delay observation vector based on the round-trip time delay observation set, and generating a segment-level observation operator corresponding to the switching state and the fiber segment based on the mapping matrix; introducing a pulse diffusion kernel characterization factor into the segment-level observation operator to construct an extended mapping matrix that can simultaneously describe the segment-level time delay accumulation effect and the pulse waveform diffusion effect; taking the segment-level time delay distortion as the solution objective, using the extended mapping matrix as the forward operator, and the time delay observation vector as the observation operator input, to construct a segment-level time delay inversion model.

[0041] Preferably, the measured end-to-end round-trip time delay values ​​under all switching states are arranged in a fixed order to form a time delay observation vector, the elements of which correspond to the time delay observation values ​​of each switching state. A mapping matrix is ​​used as the initial segment-level observation operator corresponding to the switching state and the fiber segment. It is assumed that the contribution of each fiber segment to the total time delay is simply superimposed and the pulse shape remains unchanged. The basic linear model is y=M*x, where x is the segment-level time delay distortion vector to be determined. However, in actual optical pulses, waveform broadening and diffusion will occur due to dispersion, scattering and other effects when propagating in the optical fiber. The total optical path corresponding to different switching states will vary. Different fiber lengths and segment combinations result in varying diffusion effects on the pulse, thus affecting the accurate measurement of round-trip time. Therefore, a pulse diffusion kernel characterization factor is introduced into the segment-level observation operator for more precise modeling. A function is defined for each fiber segment to characterize the diffusion or convolution effect of the pulse waveform on the time axis after the signal passes through that segment. This constructs an extended mapping matrix that simultaneously describes the segment-level delay accumulation effect and the pulse waveform diffusion effect. Given the delay and diffusion characteristics of each segment, the pulse waveform received end-to-end in any switching state can be calculated, allowing the deduction of the observed round-trip time. Finally, the segment-level delay distortion is used as the solution objective, with the extended mapping matrix as the forward operator. This forward operator includes the delay superposition information of each fiber segment and the pulse diffusion effect. The actual measured delay observation vector is used as the input to the observation operator to construct a segment-level delay inversion model. Ultimately, the segment-level delay distortion distribution that best explains the delay observation data in all switching states is determined, thereby improving the detection sensitivity and positioning accuracy for dispersion abrupt changes and nonlinear effects.

[0042] Furthermore, the specific configuration of the second identification module 30 also includes performing amplitude-phase joint consistency processing on the amplitude sequence and phase sequence of the end-to-end received signal, constructing an amplitude-phase coupled vector sequence through phase expansion, amplitude normalization and group delay compensation; and calculating the distribution divergence distance, phase deviation distance and complex domain energy distribution distance at multiple scales based on the amplitude-phase coupled vector sequence and the reference sequence, and combining the multi-scale distances to construct a switch-state statistical distance sequence.

[0043] Preferably, amplitude and phase sequences of the end-to-end received signals are subjected to amplitude-phase joint unification processing. Specifically, since the directly measured phase is usually wrapped within the principal value range, a continuous, non-jumping true phase sequence is recovered through phase expansion. Then, the amplitude sequence is normalized to eliminate the overall amplitude scaling effect caused by laser power fluctuations, changes in total link loss, etc., for example, normalizing it to a fixed total energy or peak value, so that the comparison focuses on the relative changes in the amplitude distribution shape. Then, group delay compensation is performed to calculate and compensate for the inherent group delay of the signal transmission in the link, so as to ensure that the signals of different switching states are aligned on the time axis or frequency axis, avoiding the phase linear slope and amplitude time shift introduced by different propagation times. Finally, an amplitude-phase coupling vector sequence is constructed to completely preserve the inherent coupling relationship between amplitude and phase.

[0044] Preferably, multi-scale analysis or wavelet transform is used to decompose the amplitude-phase coupling vector sequence and the reference sequence into different scales. At each scale, the distribution divergence distance, phase deviation distance, and complex domain energy distribution distance are calculated. The distribution divergence distance is calculated using KL divergence to determine the difference between the probability distribution of the amplitude at the current scale and the reference distribution at the corresponding scale, and is used to capture changes in amplitude statistical characteristics caused by loss inhomogeneity and scattering variations. The phase deviation distance is calculated to determine the statistical deviation between the unfolded phase sequence at the current scale and the reference phase sequence, and is used to capture changes in phase stability caused by polarization mode dispersion, nonlinear phase noise, etc. The complex domain energy distribution distance is calculated to determine the correlation and coherence of the energy distribution of the two complex sequences in the time-frequency domain, or the difference in their point cloud distribution on the complex plane, and is used to capture changes in the coupling relationship between amplitude and phase. For each analysis scale of each switching state, the multi-scale distances are combined in the switching state order to construct a switching state statistical distance sequence, which is used to quantify the overall deviation of the entire fiber optic link signal characteristics from the reference state at each switching state.

[0045] Furthermore, the specific configuration of the early warning module 40 also includes sending a feedback command to the endpoint control terminal if the damage confidence exceeds a preset threshold, triggering a local enhanced scan of the switch disturbance sequence. The local enhanced scan includes dynamically combining switch states in the fault abnormal segment, generating a highly focused sub-mapping matrix, and performing secondary attention acquisition analysis based on the sub-mapping matrix to establish an enhanced observation set. After performing anomaly verification of the fault abnormal segment based on the enhanced observation set, the fault abnormality is reported.

[0046] Preferably, when the damage confidence of any fiber segment exceeds a preset alarm threshold, a suspected fault is determined. First, a feedback command is sent to the endpoint control terminal, requesting a pause or temporary modification of the preset global scan sequence. This triggers a local enhancement scan of the switching disturbance sequence, dynamically generating new switching state combinations. This ensures the optical path passes through the suspected faulty segment and its adjacent area as concentrated and differentiatedly as possible, while minimizing the number of times the optical path passes through other unrelated healthy segments. Then, based on the dynamically generated combination of switching states, a highly focused sub-mapping matrix is ​​recalculated. Finally, based on the sub-mapping matrix, a secondary focus acquisition analysis is performed; that is, the endpoint control terminal drives the passive optical switch to switch to this group according to the new command. In the switch-on state, the control signal input, acquisition, time delay measurement, amplitude and phase acquisition, and complex field coherence analysis are re-executed to obtain an enhanced observation set for the suspected area. Finally, using the enhanced observation set, the anomaly verification of the faulty segment is performed through segment-level inversion. That is, the sub-mapping matrix is ​​used to invert the new time delay or amplitude and phase data. If the inversion result still clearly indicates that the segment has an anomaly and the confidence level is increased or remains high, the verification is passed, and the fault is finally confirmed. If the analysis result of the enhanced observation set shows that the anomaly is not obvious or can be attributed to noise, the damage confidence of the fiber segment is reduced, it is judged as a false alarm or temporary disturbance, no final alarm is issued, and the global scan mode may be restored.

[0047] Furthermore, the specific configuration of the early warning module 40 also includes: constructing an end-to-end residual spectrum based on historical observation sets of different time windows; performing adaptive noise floor reconstruction based on the spectral density drift of the end-to-end residual spectrum; adjusting the joint analysis weights of the first type of structured characterization and the second type of structured characterization according to the reconstructed noise floor to suppress the influence of noise floor changes on segment-level time delay inversion stability and amplitude-phase statistical distance; and performing damage analysis under multi-source observations according to the joint analysis weights.

[0048] Preferably, for each historical observation set of a historical time window, the assumed segment-level state vector is used to calculate the predicted end-to-end observations through an extended mapping matrix. Then, the actual historical observations are subtracted from the predicted observations to obtain the residual sequence, which mainly contains noise, errors, and unmodeled dynamic disturbances not explained by the model. The residual sequence is subjected to spectral analysis to determine its energy distribution at different frequency components, i.e., the end-to-end residual spectrum. Then, the residual spectrum of the recent time window is compared with the residual spectrum of the long-term historical baseline to calculate the drift of the spectral density in key frequency bands. For example, the low frequency band corresponds to slow temperature drift, and the high frequency band corresponds to random noise. Then, the noise floor reflecting the latest noise environment is dynamically reconstructed to characterize the noise level of different frequency bands.

[0049] Preferably, the joint analysis weights of the first and second type of structured representations are adjusted according to the reconstructed noise floor to suppress the impact of noise floor changes on the stability of segment-level time delay inversion and amplitude-phase statistical distance. If the reconstructed noise floor shows an increase in time delay observation noise, the weight of the time delay inversion result in the current analysis is reduced, i.e., the analysis weight of the first type of structured representation is reduced. Increased phase noise may directly lead to an increase in the benchmark value of phase deviation distance. The statistical distance threshold in the calculation of the second type of representation is dynamically adjusted or normalized and compensated according to the noise floor, and its analysis weight in the fusion decision is adjusted accordingly. Then, the first and second type of structured representations are weighted and fused using the joint analysis weights adaptively adjusted according to the noise floor, and damage analysis under multi-source observation is performed to ensure that the final fault judgment takes into account the uncertainty of the current measurement environment, thereby effectively suppressing the risk of misjudgment caused by noise changes.

[0050] Furthermore, the fiber optic end-to-end coverage fault location system based on cascaded passive optical switches also includes a fault isolation module, used to reconfigure the switching on / off state of the passive optical switches according to the fault anomaly, and to perform fault isolation management.

[0051] Preferably, the fault isolation module receives and parses faulty data to determine the target fiber segment that needs to be isolated. Based on the network topology and mapping matrix, it calculates and determines a set of reconfiguration instructions for passive optical switches, dynamically reconstructs the optical path at the physical layer, and ensures that the optical paths of all normal service signals completely bypass the identified faulty fiber segment by changing the state of one or more passive optical switches. Then, the generated switch control instructions are sent to the endpoint control terminal, which drives the corresponding passive optical switches to change their physical state. For example, if a fault occurs, the signal is switched to a pre-deployed backup fiber path or logically removed directly from the ring network by controlling the switches at both ends. After the switches are reconfigured, a fast connectivity test or status scan is automatically performed to verify that the faulty segment has been successfully bypassed and the new optical path is working normally. At the same time, the faulty segment is marked as "isolated / disabled".

[0052] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A fiber optic end-to-end coverage fault location system based on cascaded passive optical switches, characterized in that, The system includes: The mapping construction module is used to deploy multiple cascaded passive optical switches in the fiber optic link to be tested, and switch the passive optical switches at the endpoint control end according to a preset perturbation sequence to establish a mapping matrix between the switch state and the fiber segment. The first identification module is used to inject time-coded probe pulses from the endpoint in each switching state to obtain the end-to-end round-trip delay observation set for different switching states, construct a segment-level delay inversion model based on the mapping matrix, solve the local delay distortion of each fiber segment, and construct the first type of structured characterization. The second identification module is used to synchronously acquire the amplitude and phase sequences of the end-to-end received signals under different switching states, calculate the statistical distance between the synchronous acquisition results and the reference sequence, form a switching state statistical distance sequence, and project the switching state statistical distance sequence to the segment-level space according to the mapping matrix to construct the second type of structured representation. The early warning module is used to perform damage analysis based on the first type of structured characterization and the second type of structured characterization under multi-source observation, and to report fault anomalies.

2. The fiber optic end-to-end coverage fault location system based on passive optical switch cascading as described in claim 1, characterized in that, The system also includes: The third identification module is used to perform complex field detection on the end-to-end received signal under each switching state, construct a complex envelope sequence, construct a complex field cross-coherence function for any two switching states based on the complex envelope sequence, perform time-frequency analysis, extract coherent texture map and calculate texture degradation index, and construct a third type of structured representation after projection based on the mapping matrix. The location verification module is used to perform fault trust authentication based on the third type of structured representation, reconstruct the fault level based on the fault trust authentication result, and report the fault location.

3. The fiber optic end-to-end coverage fault location system based on passive optical switch cascading as described in claim 2, characterized in that, The third identification module is also used for: Gated coherent differential processing is performed based on the complex envelope sequence. In each switching state, the complex field phase of the reference state is used as the coherent anchor point. The time domain and frequency domain are jointly differentially processed through a gated window to obtain a differential coherent field that suppresses the common factor noise of the entire link. The differential coherent field is input to a multi-scale complex wavelet packet decomposer to perform amplitude-phase coupled eigenvalue decomposition on the complex field perturbation of different switching states at multiple scales, and extract scale feature clusters that reflect the perturbation response of the scatterer within the fiber segment. Based on the mapping matrix between scale feature clusters and switching states, an intra-segment perturbation fingerprint matrix is ​​constructed, and sparse inversion is performed based on spectral domain reversible reconstruction to obtain the perturbation-dominant scale and effect weight within each fiber segment, generating deep coherent texture features. The deep coherent texture features are output as an enhancement term of the third type of structured representation.

4. The fiber optic end-to-end coverage fault location system based on passive optical switch cascading as described in claim 1, characterized in that, In the first identification module, constructing the segment-level time delay inversion model includes: Based on the round-trip delay observation set, a delay observation vector is constructed, and based on the mapping matrix, a segment-level observation operator corresponding to the switching state and the fiber segment is generated; A pulse diffusion kernel characterization factor is introduced into the segment-level observation operator to construct an extended mapping matrix that can simultaneously describe the segment-level time delay accumulation effect and the pulse waveform diffusion effect; By taking the segment-level time delay distortion as the solution objective, using the extended mapping matrix as the forward operator, and the time delay observation vector as the observation operator input, a segment-level time delay inversion model is constructed.

5. The fiber optic end-to-end coverage fault location system based on passive optical switch cascading as described in claim 4, characterized in that, The second identification module is used for: Amplitude-phase joint unification processing is performed on the amplitude and phase sequences of the end-to-end received signals. An amplitude-phase coupled vector sequence is constructed through phase expansion, amplitude normalization, and group delay compensation. Based on the amplitude-phase coupling vector sequence and the reference sequence, the distribution divergence distance, phase deviation distance and complex domain energy distribution distance are calculated at multiple scales respectively, and the multi-scale distances are combined to construct a switch-state statistical distance sequence.

6. The fiber optic end-to-end coverage fault location system based on passive optical switch cascading as described in claim 1, characterized in that, The early warning module reports the following faults / abnormalities: If the damage confidence exceeds a preset threshold, a feedback command is sent to the endpoint control terminal to trigger a local enhancement scan of the switching disturbance sequence. The local enhancement scan includes dynamically combining switching states in the fault abnormal segment, generating a highly focused sub-mapping matrix, and performing secondary focus acquisition analysis based on the sub-mapping matrix to establish an enhanced observation set. After performing anomaly verification on the fault anomaly segment based on the enhanced observation set, a fault anomaly is reported.

7. The fiber optic end-to-end coverage fault location system based on passive optical switch cascading as described in claim 1, characterized in that, The early warning module is also used for: An end-to-end residual spectrum is constructed based on historical observation sets at different time windows, and an adaptive noise-low reconstruction is performed based on the spectral density drift of the end-to-end residual spectrum. The joint analysis weights of the first and second type of structured characterizations are adjusted based on the reconstructed noise floor to suppress the impact of noise floor variations on segment-level time delay inversion stability and amplitude-phase statistical distance. Damage analysis under multi-source observation is then performed based on the joint analysis weights.

8. The fiber optic end-to-end coverage fault location system based on passive optical switch cascading as described in claim 1, characterized in that, The system also includes: The fault isolation module is used to reconfigure the on / off state of the passive optical switch according to the fault anomaly, and to perform fault isolation management.

Images