Precise and synchronous collaborative data signal analysis method based on two-channel optical fiber sensing

Abstract

The invention discloses a precise and synchronous collaborative data signal analysis method based on dual-channel optical fiber sensing, and relates to the field of collaborative data signal analysis, polarization compensation and nonlinear noise decoupling analysis are carried out on time domain optical signals collected by independent A and B sensing channels to obtain polarization dimension-increasing coherent collaborative signals, and the collaborative data signal analysis is carried out on the time domain optical signals. Performing quaternion mapping and transformation compensation analysis to obtain a pure polarization state of strain immunization, performing fractional differential, HOSVD decomposition and time varying Jacobian matrix propagation disturbance based on the pure polarization state of strain immunization to obtain a non-equilibrium entanglement feature vector, combining quantum Bayesian update and a KLD weight model, generating a particle set through state sampling, and obtaining a non-equilibrium entanglement feature vector. And performing directional transfer entropy calculation on the particle set, constructing a weighted Dirichlet clustering model and directional causal energy flow analysis, comparing dual-channel energy values to obtain a channel fault judgment result, and performing channel fault judgment timely and accurately.

Description

Technical Field

[0001] The present invention relates to the field of collaborative data signal analysis, and specifically to a collaborative data signal analysis method based on precise synchronization of dual-channel optical fiber sensing. Background Art

[0002] In the field of fiber optic sensing, dual-channel synchronous monitoring technology has become a core solution for high-precision measurement in complex environments due to its ability to achieve mutual reference and redundant verification. However, existing methods have the following drawbacks: 1. Sensitive to synchronization errors: Fiber optic signals are easily affected by temperature, stress, and other factors, leading to dual-channel delay / frequency mismatch. Existing cross-correlation algorithms are limited by the sampling rate, making it difficult to achieve sub-sampling-level synchronization compensation.

[0003] 2. Polarization state interference: The optical fiber birefringence effect makes the polarization state susceptible to strain coupling interference. Traditional polarization mode dispersion compensation methods cannot dynamically decouple strain-induced polarization rotation from the actual signal characteristics.

[0004] 3. Nonlinear noise coupling: Nonlinear noise such as Rayleigh scattering and Kerr effect forms spatiotemporal coupling between the two channels, making it difficult for traditional filtering algorithms to separate multi-order noise components.

[0005] 4. The problem of decoupling multiple faults: Concurrent faults (such as strain and temperature disturbances) lead to feature aliasing. Traditional clustering algorithms rely on manual thresholds and cannot adaptively separate multimodal signals.

[0006] In order to solve the above-mentioned defects, a technical solution is now provided. Summary of the Invention

[0007] In order to solve the technical problems raised by the above background technology, the present invention is proposed. The embodiments of the present invention provide a precise and synchronized collaborative data signal analysis method based on dual-channel optical fiber sensing.

[0008] The object of the present invention can be achieved by the following technical solution: a method for analyzing collaborative data signals based on precise synchronization of dual-channel optical fiber sensing, comprising the following steps: Step S100: performing polarization compensation and nonlinear noise decoupling analysis on the time-domain optical signals collected by the independent A and B sensing channels to obtain a polarization-dimensional coherent cooperative signal; Step S200: performing quaternion mapping and transformation compensation analysis based on the polarization-dimensionality-increased coherent cooperative signal to obtain a strain-immune pure polarization state; Step S300: performing fractional differentiation, HOSVD decomposition, and propagation of disturbances in the time-varying Jacobian matrix based on the strain-immune pure polarization state to obtain a non-equilibrium entangled eigenvector; Step S400: constructing a non-homogeneous Markov transfer field based on the non-equilibrium entangled eigenvector, combining quantum Bayesian updating with the KLD weight model, and generating a particle set through state sampling; Step S500: performing directed transfer entropy calculation on the particle set, constructing a weighted Dirichlet clustering model and directed causal energy flow analysis, and comparing dual-channel energy values ​​to obtain a channel fault determination result.

[0009] Furthermore, the polarization-dimensionality-increased coherent collaborative signal analysis steps are as follows: The time domain optical signals collected by independent A and B sensing channels are marked as original signals GA(t) and GB(t). The original signals are subjected to quantum optimized mutual fuzzy matching through the quantum particle swarm mutual fuzzy model to obtain the sub-sampling level delay compensation τ0 and the sub-sampling level frequency offset correction ν0; The signal GB(t) is phase reconstructed by the polarization tensor compensation function to obtain the polarization distortion operator. Based on the subsampling level delay compensation τ0, the subsampling level frequency offset correction ν0 and the polarization distortion operator, a time-complementary polarization coherent alignment signal is generated through frequency domain phase adjustment and time offset compensation. ; Signal alignment based on time-complementary polarization coherence The sum signal GA(t) uses the tensor ring-Volterra kernel decomposition model to perform nonlinear noise decoupling and multi-dimensional feature enhancement to obtain the polarization-increased coherent cooperative signal. and .

[0010] Furthermore, the steps for analyzing the pure polarization state of the strain immunity are as follows: The Lie algebra generator sequence is constructed through the polarization rotation matrix and the Lie group logarithmic mapping is performed to obtain the strain-induced instantaneous angular velocity tensor. The strain tensor field is generated by the Lie group product integral of the strain-induced instantaneous angular velocity tensor. Based on the strain tensor field, the rotation matrix of the strain tensor is obtained through Lie algebraic exponential mapping. The inverse matrix of the rotation matrix of the strain tensor is fused with the Gaussian window function through tensor product to construct a rotation compensation operator. For the polarization discrete quaternion sequence and the rotation compensation operator, a decoupled polarization state field is established through Lie group convolution to obtain a strain-immune pure polarization state.

[0011] Furthermore, the polarization rotation matrix analysis steps are as follows: Polarization-based coherent cooperative signal and Constructing a quaternion polarization state field model, performing polarization characteristic decomposition on the quaternion polarization state field model, and obtaining a polarization discrete quaternion sequence; Directly mapping the polarization discrete quaternion sequence through the quaternion polarization field model to obtain a mapped polarization discrete quaternion sequence; Based on the mapped polarization discrete quaternion sequence, each quaternion is converted into a three-dimensional rotation group element through quaternion-rotation matrix mapping to obtain a polarization rotation matrix.

[0012] Furthermore, the non-equilibrium entanglement eigenvector analysis steps are as follows: The singular values, the time-varying Jacobian matrix components are obtained by weighted summing of the spatiotemporal main modal eigenvectors of channels A and B through outer products, and the time-varying Jacobian matrix is ​​obtained by assembling the components; Based on the phase-space entangled state vectors and time-varying Jacobian matrices of channels A and B, a tensor-driven dynamics equation is established to propagate disturbances. The fourth-order Runge-Kutta method is used to numerically integrate the equation within the adaptive cutoff time to obtain the disturbance amplification state of channel B. Based on the initial perturbation vector of the phase space of channel A and the perturbation amplification state of channel B, the polarization chaos entanglement index is calculated; The polarization chaos entanglement index and the spatiotemporal entangled high-dimensional gradient tensor are geometrically fused to obtain the non-equilibrium entanglement eigenvector.

[0013] Furthermore, the steps of analyzing the singular values ​​and the spatiotemporal main modal eigenvectors of channels A and B are as follows: Pure polarization state that is strain-immune Extract the instantaneous phase of channels A and B, and obtain the high-dimensional feature tensor of spatiotemporal entanglement based on the instantaneous phase of channels A and B through the third-order mutual information tensor model. ; High-dimensional feature tensor of space-time entanglement Using fractional-order covariant differentiators, we obtain a spatiotemporal entangled high-dimensional gradient tensor. Extract the dual-channel joint phase space trajectory from the spatiotemporal entangled high-dimensional gradient tensor and obtain the phase spatiotemporal entangled state vectors of channels A and B; Inject infinitesimal perturbations into the space-time entangled state vector of channel A phase to obtain the initial perturbation vector of channel A phase space; HOSVD decomposition is performed based on the spatiotemporal entangled high-dimensional gradient tensor to obtain the singular values ​​and the spatiotemporal main modal eigenvectors of channels A and B.

[0014] Furthermore, the particle set analysis steps are as follows: The updated posterior density matrix is ​​used to calculate the particle weights through the KLD weight update model for KL divergence and entropy change temperature to obtain the weighted particle set ; The discrete state space is determined according to the dimension of the three-dimensional transfer tensor S. The non-homogeneous space-time state transition probability is extracted based on the three-dimensional transfer tensor S. The particle state is sampled from the uniform distribution at the initial time t=0. At each time t>0, the previous state of the particle is analyzed and obtained. Based on the non-homogeneous space-time state transition probability and the previous state of the particle, the current particle state is updated by classified distribution sampling to obtain the particle state. ; Based on weighted particle set and particle states , forming a collection of particles .

[0015] Furthermore, the updated posterior density matrix analysis steps are as follows: Based on the non-equilibrium entangled eigenvectors and the generated non-equilibrium entangled characteristic sequence, a non-homogeneous Markov transfer field is constructed, including calculating the Mahalanobis distance between the non-equilibrium entangled eigenvectors at adjacent moments, and generating the time-varying transfer probability in exponential form, and outputting the three-dimensional transfer tensor S; Pure polarization state based on strain immunity Extract the instantaneous phases of channels A and B, marked as the original phase observations OA and OB of dual-channel fiber optic sensing, define the prior density ρt, the matrix of the current state probability distribution, which is initially set to uniform distribution; The probability of transferring from the previous state to the current state is extracted based on the three-dimensional transfer tensor S. The observation likelihood function and the transfer probability are combined through the quantum Bayesian updater to obtain the enhanced likelihood probability. Based on the enhanced likelihood probability, a dual-channel measurement operator is constructed. The prior states are jointly updated to obtain the updated posterior density matrix.

[0016] Furthermore, the channel fault determination result analysis steps are as follows: Based on particle state Calculate the directional transfer entropy function to obtain the causal strength from channel A to channel B; Based on the particle set, a weighted Dirichlet clustering model is used to generate a subset of failure modes. The model includes ,in Indicates dynamic threshold =0.7×max k (state confidence), where (state confidence) = , k is the discrete state number, represents clustering, Represents the set constructor symbol, and obtains the fault mode subset Cfault= ; Based on the fault mode subset, the causal strength from channel A to channel B and the particle set, the directed causal energy values ​​from channel A to channel B and the directed causal energy values ​​from channel B to channel A are calculated through the directed causal energy flow function, and channel fault judgment is performed based on the directed causal energy values.

[0017] Furthermore, the steps of analyzing the quantum particle swarm mutual fuzzy model are as follows: The quantum particle swarm mutual fuzzy model includes: , where τ is the delay search range, ν is the frequency offset search range, represents the complex conjugate of GB(t+τ), represents the frequency domain phase adjustment factor, Represents the mutual ambiguity function, i is an imaginary unit, and the quantum particle swarm algorithm is used to search for the global optimal solution, where the optimal delay and frequency offset with the largest peak value of Γ(τ, ν) are obtained, and the subsampling level delay compensation τ0 and the subsampling level frequency offset correction ν0 are obtained.

[0018] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention performs polarization compensation and nonlinear noise decoupling analysis on time-domain optical signals collected by independent A and B sensing channels to generate polarization-enhanced coherent cooperative signals. Quaternion mapping and transformation compensation analysis are then performed on the polarization-enhanced coherent cooperative signals to obtain strain-immune pure polarization states. Fractional-order differentiation, HOSVD decomposition, and time-varying Jacobian matrix propagation perturbations are then performed on these strain-immune pure polarization states to obtain non-equilibrium entangled eigenvectors. Based on these non-equilibrium entangled eigenvectors, a nonhomogeneous Markov transfer field is constructed. Combining quantum Bayesian updating with a KLD weight model, particle ensembles are generated through state sampling. This approach addresses the dual-channel delay / frequency mismatch caused by temperature and stress interference in optical fiber signals and enables subsampling-level synchronous compensation. Fiber birefringence makes polarization states susceptible to strain coupling interference. This invention dynamically decouples strain-induced polarization rotation from true signal characteristics. Nonlinear noise such as Rayleigh scattering and the Kerr effect creates spatiotemporal coupling between the two channels, and this invention can separate multi-order noise components.

[0019] 2. The present invention calculates the directional transfer entropy of particle sets, constructs a weighted Dirichlet clustering model and directional causal energy flow analysis, and compares the dual-channel energy values ​​to obtain channel fault judgment results. It can cope with concurrent faults (such as strain and temperature disturbances), adaptively separate multimodal signals, perform dynamic threshold judgment, and make channel fault judgments in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. The following drawings are not intentionally scaled to the actual size, and the focus is on illustrating the main purpose of the present invention.

[0021] Figure 1 is a flow chart of the method of the present invention; Figure 2 Flowchart of steps S100-S200 of the present invention; Figure 3 This is a flow chart of step S300 of the present invention. DETAILED DESCRIPTION

[0022] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts also fall within the scope of protection of the present invention.

[0023] like Figure 1 As shown, the precise synchronous collaborative data signal analysis method based on dual-channel optical fiber sensing includes the following steps: Step S100: performing polarization compensation and nonlinear noise decoupling analysis on the time-domain optical signals collected by the independent A and B sensing channels to obtain a polarization-dimensional coherent cooperative signal; Step S200: performing quaternion mapping and transformation compensation analysis based on the polarization-dimensionality-increased coherent cooperative signal to obtain a strain-immune pure polarization state; Step S300: performing fractional differentiation, HOSVD decomposition, and time-varying Jacobian matrix propagation disturbance based on the strain-immune pure polarization state to obtain the non-equilibrium entangled eigenvector; Step S400: constructing a non-homogeneous Markov transfer field based on the non-equilibrium entangled eigenvector, combining quantum Bayesian updating with the KLD weight model, and generating a particle set through state sampling; Step S500: performing directed transfer entropy calculation on the particle set, constructing a weighted Dirichlet clustering model and directed causal energy flow analysis, and comparing dual-channel energy values ​​to obtain a channel fault determination result.

[0024] Specifically, such as Figure 2 As shown, the analysis steps of step S100 are as follows: The time domain optical signals collected by independent A and B sensing channels are marked as original signals GA(t) and GB(t). The original signals are subjected to quantum optimized mutual fuzzy matching through the quantum particle swarm mutual fuzzy model to obtain the sub-sampling level delay compensation τ0 and the sub-sampling level frequency offset correction ν0; The signal GB(t) is phase reconstructed by the polarization tensor compensation function to obtain the polarization distortion operator. Based on the subsampling level delay compensation τ0, the subsampling level frequency offset correction ν0 and the polarization distortion operator, a time-complementary polarization coherent alignment signal is generated through frequency domain phase adjustment and time offset compensation. ; Signal alignment based on time-complementary polarization coherence The sum signal GA(t) uses the tensor ring-Volterra kernel decomposition model to perform nonlinear noise decoupling and multi-dimensional feature enhancement to obtain the polarization-increased coherent cooperative signal. and ; Specifically, the time domain optical signals collected by two independent sensing channels A and B are marked as original signals GA(t) and GB(t). The original signals GA(t) and GB(t) are subjected to quantum optimized mutual fuzzy matching through a quantum particle swarm mutual fuzzy model to obtain the subsampling level time delay compensation τ0 and the subsampling level frequency offset correction ν0. The model includes: τ is the delay search range, which represents the time difference between the two channel signals; ν is the frequency offset search range, which represents the frequency difference between the two channel signals. Represents the complex conjugate of GB(t+τ), which is used to calculate the phase information of the cross-correlation and extract the joint features of time delay and frequency offset. Represents the frequency domain phase adjustment factor, which compensates for the frequency offset and aligns the two channel signals in the frequency domain. represents the mutual ambiguity function, i is an imaginary unit, and the quantum particle swarm algorithm is used to search for the global optimal solution, so as to obtain the optimal delay and frequency offset that maximizes the peak value of Γ(τ, ν), and obtain the subsampling level delay compensation τ0 and the subsampling level frequency offset correction ν0; The phase of the signal GB(t) is reconstructed by the polarization tensor compensation function to obtain the polarization distortion operator , where LA and LB are dual-channel polarization state matrices, represents the Kronecker product, represents conjugate transpose, det represents the determinant of the matrix, reflecting the linear transformation characteristics of the matrix, Represents the phase angle. Based on the subsampling level delay compensation τ0, the subsampling level frequency offset correction ν0 and the polarization distortion operator, a time-compensated polarization coherent alignment signal is generated through frequency domain phase adjustment and time offset compensation. , specifically, , where F(*) represents Fourier transform, represents the inverse transform of Fourier transform, the frequency offset correction ν0 and polarization phase DP at the subsampling level, and then converted back to the time domain, (t-τ0) represents the time domain delay correction, represents the frequency domain compensation factor; Signal alignment based on time-complementary polarization coherence The sum signal GA(t) uses the tensor ring-Volterra kernel decomposition model to perform nonlinear noise decoupling and multi-dimensional feature enhancement to obtain the polarization-increased coherent cooperative signal. and , the model includes: 𝑆 where ∏ represents the product operation, k represents the order of the Volterra series, 1st order is linear noise, 2nd order is quadratic nonlinear noise, and 3rd order is cubic nonlinear noise, covering the complex noise model in fiber optic sensing. Represents the tensor ring decomposition kernel, TR-Volterra kernel, which describes the spatiotemporal response characteristics of noise of various orders through tensor decomposition optimization. The dimension matches the signal tensor. G is the dual-channel synchronization signal vector. , T represents the transpose of the matrix, represents the time delay variable, the time offset in the integration domain, capturing the memory effect of the signal, such as the multipath delay of scattering noise, j is the index of the continuous product operation, used to traverse the time delay component in the k-order nonlinear term, Represents the Volterra series model, which outputs polarization-increased coherent cooperative signals and ; Specifically, the analysis steps in step S200 are as follows: Polarization-based coherent cooperative signal and Constructing a quaternion polarization state field model, performing polarization characteristic decomposition on the quaternion polarization state field model, and obtaining a polarization discrete quaternion sequence; Directly mapping the polarization discrete quaternion sequence through the quaternion polarization field model to obtain a mapped polarization discrete quaternion sequence; Based on the mapped polarization discrete quaternion sequence, each quaternion is converted into a three-dimensional rotation group element through quaternion-rotation matrix mapping to obtain a polarization rotation matrix; The Lie algebra generator sequence is constructed through the polarization rotation matrix and the Lie group logarithmic mapping is performed to obtain the strain-induced instantaneous angular velocity tensor. The strain tensor field is generated by the Lie group product integral of the strain-induced instantaneous angular velocity tensor. Based on the strain tensor field, the rotation matrix of the strain tensor is obtained through Lie algebraic exponential mapping. The inverse matrix of the rotation matrix of the strain tensor is fused with the Gaussian window function through tensor product to construct a rotation compensation operator. For the polarization discrete quaternion sequence and the rotation compensation operator, a decoupled polarization state field is established through Lie group convolution to obtain a pure polarization state that is strain-immune. Specifically, polarization-dimensional coherent cooperative signal and Construct a quaternion polarization state field model, which includes: , where Re(*) represents the real part of the complex number, Im(*) represents the imaginary part of the complex number, i, j, and k represent the quaternion imaginary unit, describing the three-dimensional polarization space rotation and coupling, satisfying the quaternion algebraic rules, W(t) represents the quaternion signal field, quantizing the spatiotemporal evolution of the dual-channel polarization state, and obtaining the polarization discrete quaternion sequence , n represents the time sequence number, the maximum value is N; Polarization discrete quaternion sequence , directly mapped through the quaternion polarization field model , get the mapped polarization discrete quaternion sequence , where fr and fj are the real parts of the polarization-dimensional coherent cooperation signal, and fi and fk are the imaginary parts of the polarization-dimensional coherent cooperation signal; Based on the mapped polarization discrete quaternion sequence, each quaternion is converted into a three-dimensional rotation group element through quaternion-rotation matrix mapping to obtain the polarization rotation matrix , specifically:

[0025] By polarization rotation matrix A Lie algebra generator sequence is constructed to perform Lie group logarithmic mapping to obtain the strain-induced instantaneous angular velocity tensor. The Lie algebra generator sequence is: , where D m The strain tensor field is generated by Lie group product integration of the instantaneous angular velocity tensor induced by strain. , specific , where exp (*) represents the exponential mapping, ∏ represents the Lie group product integral, multiple moment rotation accumulation, generates the strain tensor, Log is the cumulative strain tensor, and its Frobenius norm is quantify strain intensity; Based on the strain tensor field Through Lie algebraic index mapping, the rotation matrix of the strain tensor is obtained , , is the antisymmetric component of the strain tensor field, , exp(*) is the Lie group exponential mapping, which converts the shear strain into polarization state rotation based on the inverse matrix of the rotation matrix of the strain tensor Perform tensor product fusion with Gaussian window function to construct rotation compensation operator J(t). , is the Gaussian window function, is the signal coherence time, is the tensor product (which realizes the joint operation of time domain filtering and spatial rotation); Polarization discrete quaternion sequence And the rotation compensation operator J(t), through the Lie group convolution to establish the decoupled polarization state field, to obtain the strain-immune pure polarization state , ,in is the integration dummy variable, which represents the offset within the time sliding window; Specifically, such as Figure 3 As shown, the analysis steps of step S300 are as follows: Pure polarization state that is strain-immune Extract the instantaneous phase of channels A and B, and obtain the high-dimensional feature tensor of spatiotemporal entanglement based on the instantaneous phase of channels A and B through the third-order mutual information tensor model. ; High-dimensional feature tensor of space-time entanglement Using fractional-order covariant differentiators, we obtain a spatiotemporal entangled high-dimensional gradient tensor. Extract the dual-channel joint phase space trajectory from the spatiotemporal entangled high-dimensional gradient tensor and obtain the phase spatiotemporal entangled state vectors of channels A and B; Inject infinitesimal perturbations into the space-time entangled state vector of channel A phase to obtain the initial perturbation vector of channel A phase space; HOSVD decomposition is performed based on the spatiotemporal entangled high-dimensional gradient tensor to obtain the singular values ​​and spatiotemporal main modal eigenvectors of channels A and B; The singular values, the time-varying Jacobian matrix components are obtained by weighted summing of the spatiotemporal main modal eigenvectors of channels A and B through outer products, and the time-varying Jacobian matrix is ​​obtained by assembling the components; Based on the phase-space entangled state vectors and time-varying Jacobian matrices of channels A and B, a tensor-driven dynamics equation is established to propagate disturbances. The fourth-order Runge-Kutta method is used to numerically integrate the equation within the adaptive cutoff time to obtain the disturbance amplification state of channel B. Based on the initial perturbation vector of the phase space of channel A and the perturbation amplification state of channel B, the polarization chaos entanglement index is calculated; The polarization chaos entanglement index and the spatiotemporal entangled high-dimensional gradient tensor are geometrically fused to obtain the non-equilibrium entanglement eigenvector.

[0026] Specifically, a pure polarization state that is immune to strain Extract the instantaneous phase of channels A and B , , arg represents the phase angle of the complex number to be extracted, 、 Represent the instantaneous phase of channels A and B respectively, and obtain the high-dimensional feature tensor of spatiotemporal entanglement through the third-order mutual information tensor model. , the model includes: , where I is the mutual information of three variables, I(X;Y;Z)=H(X)+H(Y)+H(Z)-H(X,Y)-H(X,Z)-H(Y,Z)+H(X,Y,Z), H is the information entropy, X, Y, Z are the inputs of the mutual information of three variables, is the time lag parameter; High-dimensional feature tensor of space-time entanglement Using fractional-order covariant differentiators, we can obtain the spatiotemporal entangled high-dimensional gradient tensor. Specifically, , represents the spatiotemporal entangled high-dimensional gradient tensor, α is the fractional order, is the Gamma function, which extracts the long-term memory characteristics of the signal. δ represents the historical time variable. Represents the high-dimensional feature tensor of spatiotemporal entanglement; The dual-channel joint phase space trajectory is extracted from the spatiotemporal entangled high-dimensional gradient tensor to obtain the phase spatiotemporal entangled state vectors of channels A and B. Specifically: , The first dimension index of the fixed spatiotemporal entangled high-dimensional gradient tensor is i (channel A at time ti), the third dimension index is k1 (specific time lag Δtk1), and the fractional mutual information gradient distribution of channel B at all times is, The second dimension index of the fixed spatiotemporal entangled high-dimensional gradient tensor is j (channel B at time tj), the third dimension index is k1 (specific time lag Δtk1), and the fractional mutual information gradient distribution of channel A at all times is, 、 Represents the phase-time entangled state vector of channels A and B; Inject infinitesimal perturbations into the space-time entangled state vector of channel A to obtain the initial perturbation vector of channel A phase space. Specifically, , represents the initial perturbation vector of the phase space of channel A, represents the initial moment, u represents the perturbation of the minimum eigenvector of the spatiotemporal entanglement state vector along channel A, ,in represents a unit vector, Represents the state vector along The gradient of the direction, arg min represents the search for the independent variable value that makes the function achieve the minimum value, Represents the Frobenius norm operation; Based on the spatiotemporal entangled high-dimensional gradient tensor, HOSVD decomposition is performed to obtain the singular value σ l, the spatiotemporal main modal eigenvectors of channels A and B and ; Specifically, HOSVD decomposition reduces the dimensionality of the spatiotemporal entangled high-dimensional gradient tensor, extracts the main modal features, separates the core spatiotemporal coupling information, removes noise, simplifies calculations, and provides low-dimensional high-resolution features for subsequent chaos analysis, anomaly detection, and strain decoupling, thereby improving sensing accuracy and robustness. The singular values, the time-varying Jacobian matrix components are obtained by weighted summing of the outer products of the main modal eigenvectors of channels A and B, and the time-varying Jacobian matrix is ​​obtained by assembling the components. Specifically: , where l is the sum index variable, indicating the modal number currently involved in the calculation, and r is the cutoff rank, indicating the total number of modes retained. represents the outer product operation, constructs the cross-channel correlation matrix, and combines with other components to form the time-varying Jacobian matrix; Based on the phase-time entangled state vectors and time-varying Jacobian matrices of channels A and B, a tensor-driven dynamics equation is established to propagate disturbances, where the tensor-driven dynamics equation is: ,in 、 denote the phase space-time entangled state disturbance vectors of channels A and B respectively, Denotes the Jacobian matrix, and the fourth-order Runge-Kutta method is used to numerically integrate the equation within the adaptive cutoff time Tc=2π / ωdom to obtain the disturbance amplification state of channel B , ωdom is the dominant frequency of the power spectrum of the spatiotemporal entangled high-dimensional gradient tensor; Based on the initial perturbation vector of the phase space of channel A and the perturbation amplification state of channel B Calculate the polarization chaos entanglement index, specifically, , is the polarization chaos entanglement index; The polarization chaos entanglement index and the space-time entanglement high-dimensional gradient tensor are geometrically fused to obtain the non-equilibrium entanglement characteristic vector. Specifically: , where Re represents the real part of the complex number to be extracted, ω represents the continuous weight parameter, represents the Frobenius norm, E represents the non-equilibrium entanglement eigenvector; Specifically, the analysis steps in step S400 are as follows: Based on the non-equilibrium entangled characteristic vector, the generated non-equilibrium entangled characteristic sequence , where each The topological feature encoder outputs a high-dimensional feature vector, where t represents the time variable and has a value range of 1-T, where T is the maximum value of the time variable. This constructs a non-homogeneous Markov transition field, including calculating the Mahalanobis distance between the entangled feature vectors of the non-equilibrium state at adjacent moments, and generates the time-varying transition probability in exponential form, outputting a three-dimensional transition tensor S. The above non-homogeneous Markov transition field includes the formula: ,in Represents the temperature parameter, which controls the probability decay rate, i and j represent the state at time t, represents the inverse covariance matrix of the non-equilibrium entangled eigenvector, represents Mahalanobis distance, weighted Euclidean distance, Represents the probability of transitioning from state j to state i at time t. The dimension of the three-dimensional transition tensor S is the number of time points × the number of discrete states × the number of discrete states, which stores the state transition probabilities at all times; Pure polarization state based on strain immunity Extract the instantaneous phases of channels A and B, marked as the original phase observations OA and OB of the dual-channel fiber optic sensor, and define the prior density ρt, the matrix of the current state probability distribution, which is initially set to be uniformly distributed, ρ0=I / K, where K is the number of discrete states and I is the unit matrix, which is subsequently updated based on historical information; Based on the three-dimensional transition tensor S, the probability of the previous state transferring to the current state is extracted. ,in Represents the three-dimensional transition tensor S from the previous state at time t Transfer to current state The probability of Indicates the previous state Transfer to current state The transition probability of is combined with the observation likelihood function and the transition probability through the quantum Bayesian updater to obtain the enhanced likelihood probability: , where X represents the channel identifier, specifically: X∈{A, B}, corresponding to the dual channels, Indicates a given state Time observation , Indicates that in the current observation , the enhanced likelihood probability of the current hidden state and the previous hidden state, and construct a dual-channel measurement operator based on the enhanced likelihood probability. ,in represents the measurement operator of channel X, Represents a unitary transformation matrix, maintaining probability conservation, diag(*) represents converting the probability vector into a diagonal matrix, represents the conjugate transpose operation; The prior state is jointly updated to obtain the updated posterior density matrix, , M A 、M B represents the measurement operator for channels A and B, represents the updated posterior density matrix, Represents tensor product, fusing dual-channel measurements, represents matrix trace operation and probability normalization; The updated posterior density matrix is ​​used to calculate the particle weights through the KLD weight update model for KL divergence and entropy change temperature to obtain the weighted particle set , the specific KLD weight update model includes ,in Represents the prior density matrix of particle n, which is obtained by processing the prior density matrix through the entropy-driven particle filter. represents the weight of particle n, represents KL divergence, quantifying the difference between the posterior and the prior, Q represents von Neumann entropy, Q=-tr(ρlnρ), ρ represents the density matrix, and ∝ represents a proportional operation; According to the dimension of the three-dimensional transfer tensor S, the discrete state space S0={1, 2, ..., K} is determined, where K is the number of discrete states, and the non-homogeneous spatiotemporal state transition probability is extracted based on the three-dimensional transfer tensor S. , where n represents the particle index, k represents the target state belongs to {1, 2, ..., K}, m represents the previous state, col(m) represents the matrix column index corresponding to the state value m, Indicates that the current state value of particle n is equal to k, Indicates that the state value of particle n in the previous state is equal to m, represents the non-homogeneous space-time state transition probability, the probability of state k transferring to state col(m), and the particle state is sampled from the uniform distribution at the initial time t=0 , Indicates that the initial state obeys a uniform distribution. Each particle n represents a state hypothesis. The initial particle state is set to the maximum mixed state ρ0=I / K, where K is the number of discrete states and I is the unit matrix. At each moment t>0, according to the previous state of the particle , m represents the state index, based on the non-homogeneous space-time state transition probability and the previous state of the particle, and updates the current particle state through classified distribution sampling , Indicates that the current state of particle n obeys the category distribution, and the particle state is obtained ; Based on weighted particle set and particle states , forming a collection of particles ; Specifically, the analysis steps in step S500 are as follows: Based on particle state Calculate the directional transfer entropy function to obtain the causal strength from channel A to channel B. ,in represents the causal strength from channel A to channel B, Δ represents the time lag, represents the mutual information term in the directional transfer entropy of the channel state A at time t to the channel state B at time t+Δ, Represents the mutual information term in the directional transfer entropy of the state of channel A at time t to the state of channel B at time t-Δ, and separates the particle states through integer operations ,get and , respectively represent the state sequence of channel A at time t and the state sequence of channel B at time t, I(X;Y)=H(X)+H(Y)-H(X,Y), H is the information entropy, X and Y are the inputs of the mutual information of the two variables, It means taking the maximum value of the time lag Δ and screening the most significant causal transmission; Based on the particle set, a weighted Dirichlet clustering model is used to generate a subset of failure modes. The model includes ,in Indicates dynamic threshold =0.7×max k (state confidence), where (state confidence) = , k is the discrete state number, represents clustering, Represents a collection constructor symbol that satisfies subsequent conditions The set only contains the ones with significant weights, realizes fault mode clustering, and obtains the fault mode subset Cfault= ; Based on the fault mode subset, the causal strength from channel A to channel B, and the particle set, the directional causal energy value from channel A to channel B and the directional causal energy value from channel B to channel A are calculated through the directional causal energy flow function, and the channel fault is determined based on the directional causal energy value; Specifically, the directed causal energy flow function: ,in Represents the directional causal energy value from channel A to channel B. Similarly, the directional causal energy value from channel B to channel A is calculated. , and perform the directed causal energy value from channel A to channel B and the directional causal energy value from channel B to channel A ,like Greater than , it indicates that the fault source is in channel A, otherwise it indicates that the fault source is in channel B; Specifically, after constructing a discrete state space using particle sets, the weighted Dirichlet clustering model first calculates the state confidence corresponding to each discrete state number k, i.e., the ratio of the sum of all particle weights in that state to the total particle weights. A dynamic threshold τ, set at 70% of the maximum state confidence, is used to screen states with weights significantly higher than τ. Using the set constructor symbol, particles that meet the criteria are clustered into independent fault mode subsets Ck by state k, achieving decoupling of concurrent faults (e.g., separating different perturbations such as strain and temperature). For the target fault mode subset Cfault, the bidirectional energy integral between channels A and B is calculated in real time using a directed causal energy flow function, combining the causal strength between channels A and B. The directed causal energy from A to B is compared with that from B to A: If the energy from A to B is greater, the fault source is determined to be in channel A; otherwise, it is in channel B. This process, driven by particle weights, dynamically adapts to complex scenarios, promptly screens significant fault modes, and rapidly locates dual-channel faults based on the directionality of energy flow. It effectively handles concurrent multi-fault scenarios and provides real-time and accurate fault determination capabilities for applications such as industrial monitoring and structural health diagnosis.

[0027] Specifically, a quantum particle swarm mutual fuzzification model is first used to achieve subsampling-level delay and frequency offset correction. Nonlinear noise decoupling is achieved by combining tensor ring-Volterra kernel decomposition, generating polarization-dimensional coherent cooperative signals. A quaternion polarization state field model is then constructed, and the strain tensor field is generated using Lie group logarithmic mapping and product integral. A rotation compensation operator is constructed via Lie algebraic exponential mapping to achieve strain-immune pure polarization state extraction. The pure polarization state is then subjected to fractional covariant differentiation and HOSVD decomposition, and perturbations are propagated using a time-varying Jacobian matrix to generate non-equilibrium entangled eigenvectors. Based on these eigenvectors, a nonhomogeneous Markov transfer field is constructed. Quantum Bayesian updating is combined with the KLD weight model to generate a weighted particle set. Dual-channel fault source location is achieved through directed transfer entropy and weighted Dirichlet cluster analysis combined with a directed causal energy flow function. By deeply coupling interdisciplinary technologies such as quantum optimization, quaternion algebra, Lie group geometry and tensor decomposition, the limitation of traditional fiber optic sensing relying on a single signal dimension is broken through. Through the decoupling of spatiotemporal entanglement characteristics and non-equilibrium dynamics modeling, significant improvements in sub-sampling delay compensation accuracy, strain-immune polarization state purity and fault location accuracy are achieved, which can adapt to the precise synchronization and intelligent diagnosis of fiber optic sensing systems in complex environments.

[0028] The above is an illustration of the present invention and should not be considered as limiting thereof. Although several exemplary embodiments of the present invention have been described, it will be readily understood by those skilled in the art that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present invention. Therefore, all such modifications are intended to be included within the scope of the present invention as defined by the claims. It should be understood that the above is an illustration of the present invention and should not be considered as being limited to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present invention is defined by the claims and their equivalents.

Claims

1. A collaborative data signal analysis method based on precise synchronization of dual-channel optical fiber sensing, characterized in that: The following steps are involved: Step S100: performing polarization compensation and nonlinear noise decoupling analysis on the time-domain optical signals collected by the independent A and B sensing channels to obtain a polarization-dimensional coherent cooperative signal; Step S200: performing quaternion mapping and transformation compensation analysis based on the polarization-dimensionality-increased coherent cooperative signal to obtain a strain-immune pure polarization state; Step S300: performing fractional differentiation, HOSVD decomposition, and time-varying Jacobian matrix propagation disturbance based on the strain-immune pure polarization state to obtain a non-equilibrium entangled eigenvector; Step S400: constructing a non-homogeneous Markov transfer field based on the non-equilibrium entangled eigenvector, combining quantum Bayesian updating with the KLD weight model, and generating a particle set through state sampling; Step S500: performing directed transfer entropy calculation on the particle set, constructing a weighted Dirichlet clustering model and directed causal energy flow analysis, and comparing dual-channel energy values ​​to obtain a channel fault determination result.

2. The method for analyzing precise and synchronized collaborative data signals based on dual-channel optical fiber sensing according to claim 1, characterized in that: The steps of polarization dimension-enhanced coherent collaborative signal analysis are as follows: The time-domain optical signals collected by independent A and B sensing channels are marked as original signals GA(t) and GB(t). The original signals are subjected to quantum optimized mutual fuzzy matching based on the quantum particle swarm mutual fuzzy model to obtain the sub-sampling level time delay compensation τ0 and the sub-sampling level frequency offset correction ν0. The signal GB(t) is phase reconstructed by the polarization tensor compensation function to obtain the polarization distortion operator. Based on the subsampling level delay compensation τ0, the subsampling level frequency offset correction ν0 and the polarization distortion operator, a time-complementary polarization coherent alignment signal is generated through frequency domain phase adjustment and time offset compensation. ; Signal alignment based on time-complementary polarization coherence The sum signal GA(t) uses the tensor ring-Volterra kernel decomposition model to perform nonlinear noise decoupling and multi-dimensional feature enhancement to obtain the polarization-increased coherent cooperative signal. and .

3. The method for analyzing collaborative data signals based on precise synchronization of dual-channel optical fiber sensing according to claim 1, characterized in that: The steps for strain immunity pure polarization state analysis are as follows: The Lie algebra generator sequence is constructed through the polarization rotation matrix and the Lie group logarithmic mapping is performed to obtain the strain-induced instantaneous angular velocity tensor. The strain tensor field is generated by the Lie group product integral of the strain-induced instantaneous angular velocity tensor. Based on the strain tensor field, the rotation matrix of the strain tensor is obtained through Lie algebraic exponential mapping. The inverse matrix of the rotation matrix of the strain tensor is fused with the Gaussian window function through tensor product to construct a rotation compensation operator. For the polarization discrete quaternion sequence and the rotation compensation operator, a decoupled polarization state field is established through Lie group convolution to obtain a strain-immune pure polarization state.

4. The method for analyzing precise and synchronized collaborative data signals based on dual-channel optical fiber sensing according to claim 3, characterized in that: The polarization rotation matrix analysis steps are as follows: Polarization-based coherent cooperative signal and Constructing a quaternion polarization state field model, performing polarization characteristic decomposition on the quaternion polarization state field model, and obtaining a polarization discrete quaternion sequence; Directly mapping the polarization discrete quaternion sequence through the quaternion polarization field model to obtain a mapped polarization discrete quaternion sequence; Based on the mapped polarization discrete quaternion sequence, each quaternion is converted into a three-dimensional rotation group element through quaternion-rotation matrix mapping to obtain a polarization rotation matrix.

5. The method for analyzing precise and synchronized collaborative data signals based on dual-channel optical fiber sensing according to claim 1, characterized in that: The steps for analyzing the non-equilibrium entanglement eigenvector are as follows: The singular values, the spatiotemporal main modal eigenvectors of channels A and B are weightedly summed by outer products to obtain the components of each time-varying Jacobian matrix, and the components are assembled to obtain the time-varying Jacobian matrix; Based on the phase-space entangled state vectors and time-varying Jacobian matrices of channels A and B, a tensor-driven dynamics equation is established to propagate disturbances. The fourth-order Runge-Kutta method is used to numerically integrate the equation within the adaptive cutoff time to obtain the disturbance amplification state of channel B. Based on the initial perturbation vector of the phase space of channel A and the perturbation amplification state of channel B, the polarization chaos entanglement index is calculated; The polarization chaos entanglement index and the spatiotemporal entangled high-dimensional gradient tensor are geometrically fused to obtain the non-equilibrium entanglement eigenvector.

6. The method for analyzing collaborative data signals based on precise synchronization of dual-channel optical fiber sensing according to claim 5, characterized in that: The steps for analyzing the singular values ​​and the spatiotemporal main modal eigenvectors of channels A and B are as follows: Pure polarization state that is strain-immune Extract the instantaneous phase of channels A and B, and obtain the high-dimensional feature tensor of spatiotemporal entanglement based on the instantaneous phase of channels A and B through the third-order mutual information tensor model. ; High-dimensional feature tensor of space-time entanglement Using fractional-order covariant differentiators, we obtain a spatiotemporal entangled high-dimensional gradient tensor. Extract the dual-channel joint phase space trajectory from the spatiotemporal entangled high-dimensional gradient tensor and obtain the phase spatiotemporal entangled state vectors of channels A and B; Inject infinitesimal perturbations into the space-time entangled state vector of channel A phase to obtain the initial perturbation vector of channel A phase space; HOSVD decomposition is performed based on the spatiotemporal entangled high-dimensional gradient tensor to obtain the singular values ​​and the spatiotemporal main modal eigenvectors of channels A and B.

7. The method for analyzing precise and synchronized collaborative data signals based on dual-channel optical fiber sensing according to claim 1, characterized in that: The particle set analysis steps are as follows: The updated posterior density matrix is ​​used to calculate the particle weights through the KLD weight update model for KL divergence and entropy change temperature to obtain the weighted particle set ; The discrete state space is determined according to the dimension of the three-dimensional transfer tensor S. The non-homogeneous space-time state transition probability is extracted based on the three-dimensional transfer tensor S. The particle state is sampled from the uniform distribution at the initial time t=0. At each time t>0, the previous state of the particle is analyzed and obtained. Based on the non-homogeneous space-time state transition probability and the previous state of the particle, the current particle state is updated by classified distribution sampling to obtain the particle state. ; Based on weighted particle set and particle states , forming a collection of particles .

8. The method for analyzing collaborative data signals based on precise synchronization of dual-channel optical fiber sensing according to claim 7, characterized in that: The updated posterior density matrix analysis steps are as follows: Based on the non-equilibrium entangled eigenvectors and the generated non-equilibrium entangled characteristic sequence, a non-homogeneous Markov transfer field is constructed, including calculating the Mahalanobis distance between the non-equilibrium entangled eigenvectors at adjacent moments, and generating the time-varying transfer probability in exponential form, and outputting the three-dimensional transfer tensor S; Pure polarization state based on strain immunity Extract the instantaneous phases of channels A and B, marked as the original phase observations OA and OB of dual-channel fiber optic sensing, define the prior density ρt, the matrix of the current state probability distribution, which is initially set to uniform distribution; The probability of transferring from the previous state to the current state is extracted based on the three-dimensional transfer tensor S. The observation likelihood function and the transfer probability are combined through the quantum Bayesian updater to obtain the enhanced likelihood probability. Based on the enhanced likelihood probability, a dual-channel measurement operator is constructed. The prior states are jointly updated to obtain the updated posterior density matrix.

9. The method for analyzing precise and synchronized collaborative data signals based on dual-channel optical fiber sensing according to claim 1, characterized in that: The steps for analyzing the channel fault determination results are as follows: Based on particle state Calculate the directional transfer entropy function to obtain the causal strength from channel A to channel B; Based on the particle set, a weighted Dirichlet clustering model is used to generate a subset of failure modes. The model includes ,in represents the dynamic threshold, =0.7×max k , max k is the state confidence, state confidence = , k is the discrete state number, represents clustering, Represents the set constructor symbol, and obtains the fault mode subset Cfault= ; Based on the fault mode subset, the causal strength from channel A to channel B and the particle set, the directed causal energy values ​​from channel A to channel B and the directed causal energy values ​​from channel B to channel A are calculated through the directed causal energy flow function, and channel fault judgment is performed based on the directed causal energy values.

10. The method for analyzing precise and synchronized collaborative data signals based on dual-channel optical fiber sensing according to claim 1, characterized in that: The analysis steps of the quantum particle swarm mutual fuzzy model are as follows: The quantum particle swarm mutual fuzzy model includes: , where τ is the delay search range, ν is the frequency offset search range, represents the complex conjugate of GB(t+τ), represents the frequency domain phase adjustment factor, Represents the mutual ambiguity function, i is an imaginary unit, and the quantum particle swarm algorithm is used to search for the global optimal solution, where the optimal delay and frequency offset with the largest peak value of Γ(τ, ν) are obtained, and the subsampling level delay compensation τ0 and the subsampling level frequency offset correction ν0 are obtained.

Images