An intelligent monitoring method and device based on optical fiber sensing and optical neural network

Abstract

The present application provides an intelligent monitoring method and device based on optical fiber sensing and optical neural network, which relates to the field of industrial monitoring. The method includes: converting the physical quantity change signal into a desensitized signal through the Brillouin scattering frequency shift effect; three-dimensionally encoding the desensitized signal to obtain the nonlinear mapping of the physical quantity change signal in the feature space; constructing a diffraction neural network using a preset phase-changing material; inputting the physical quantity change signal into the diffraction neural network, and the diffraction neural network implements iterative updates of the weight matrix based on the back-propagation optical path; decoupling the four-dimensional sensing quantity in the physical quantity change signal through the diffraction neural network, and analyzing and outputting the early warning signal through the micro-ring resonant cavity array. The present application solves the problem that traditional electrical sensors are easily interfered with by external electromagnetic fields in strong electromagnetic environments because their signals depend on the electrical signal transmission path, resulting in amplitude drift and frequency fluctuations, and therefore have signal distortion in complex electromagnetic environments.

Description

Technical Field

[0001] The present application relates to the field of industrial monitoring, and in particular to an intelligent monitoring method and device based on optical fiber sensing and optical neural network. Background Art

[0002] Today, the industrial monitoring field is facing a trend of increasingly complex perception environments driven by extreme working conditions such as high temperature, high pressure, high radiation and strong electromagnetic interference, which places higher demands on the sensor's signal fidelity, real-time response capability and anti-interference performance.

[0003] Traditional electrical sensors typically measure tiny deformations of external structures using strain gauges, converting these deformation signals into changes in voltage or resistance to quantify stress or displacement. However, because these signals rely on electrical signal transmission paths, they are susceptible to interference from external electromagnetic fields in strong electromagnetic environments, resulting in amplitude drift and frequency fluctuations. This is particularly noticeable in high-power devices, variable-frequency motors, or strong radar. Therefore, traditional electrical sensors suffer from signal distortion in complex electromagnetic environments.

[0004] Therefore, there is an urgent need for an intelligent monitoring method and device based on optical fiber sensing and optical neural network. Summary of the Invention

[0005] The present application provides an intelligent monitoring method and device based on optical fiber sensing and optical neural network, which solves the problem that traditional electrical sensors are easily interfered with by external electromagnetic fields in strong electromagnetic environments, resulting in amplitude drift and frequency fluctuations, and thus signal distortion in complex electromagnetic environments.

[0006] In a first aspect of the present application, an intelligent monitoring method based on optical fiber sensing and optical neural network is provided, the method comprising: collecting physical quantity change signals based on a Bragg grating array, and converting the physical quantity change signals into desensitized signals through the Brillouin scattering frequency shift effect; performing three-dimensional encoding on the desensitized signals based on a Mach-Zehnder interferometer array to obtain a nonlinear mapping of the physical quantity change signals in the feature space; constructing a diffraction neural network based on the nonlinear mapping of the feature space and using a preset phase-changing material; inputting the physical quantity change signals into the diffraction neural network, and the diffraction neural network realizing iterative updating of the weight matrix based on a back-propagation optical path; decoupling the four-dimensional sensing quantity in the physical quantity change signals through the diffraction neural network, and analyzing and outputting a warning signal through a microring resonant cavity array.

[0007] Optionally, a physical quantity change signal is collected based on a Bragg grating array, and the physical quantity change signal is converted into a desensitized signal through the Brillouin scattering frequency shift effect, specifically including: calculating the Brillouin scattering frequency shift amount through the physical quantity change signal based on the Brillouin scattering frequency shift effect of the Bragg grating array; based on the Brillouin scattering frequency shift amount, eliminating the temperature-strain cross sensitivity in the physical quantity change signal through a double-pulse difference method to convert the physical quantity change signal into a desensitized signal.

[0008] Optionally, the desensitized signal is three-dimensionally encoded based on a Mach-Zehnder interferometer array, specifically including: three-dimensionally encoding the desensitized signal based on a Mach-Zehnder interferometer array, the three-dimensional encoding is time-frequency-phase three-dimensional encoding, and the transmission matrix of the time-frequency-phase three-dimensional encoding realizes interference reconstruction between different paths by adjusting the phase difference.

[0009] Optionally, a diffraction neural network is constructed based on the nonlinear mapping of the feature space and using a preset phase-changing material, specifically including: guiding the nonlinear mapping of the feature space to an optical matrix processor, the optical matrix processor being composed of a Mach-Zehnder interferometer array; performing multi-layer unitary transformation operations based on a weight matrix through each layer of the optical matrix processor, and guiding the weight matrix to be iteratively updated based on the phase change generated in the back-propagation optical path, so as to construct a diffraction neural network using a preset phase-changing material.

[0010] Optionally, each layer of optical matrix processor receives the input light field distribution of the nth layer and processes the input light field distribution under the action of a nonlinear function;

[0011] The output light field distribution of the n+1th layer is generated through the integral mapping relationship with the weight matrix of the nth layer.

[0012] Optionally, the diffraction neural network implements iterative updating of the weight matrix based on the back-propagation optical path, specifically including: the weight matrix is ​​determined by the etching depth of the hologram.

[0013] Optionally, the four-dimensional sensing quantity in the physical quantity change signal is obtained through a diffraction neural network, specifically including: establishing a joint distribution model of the four-dimensional sensing quantity by constructing a vibration equation, a temperature-strain coupling equation and a chemical concentration kinetic equation; and decoupling the joint distribution model through tensor decomposition of the diffraction neural network.

[0014] In a second aspect of the present application, an intelligent monitoring device based on optical fiber sensing and optical neural network is provided, the device including a processing module and an early warning module, wherein:

[0015] The processing module is used to collect physical quantity change signals based on a Bragg grating array, and convert the physical quantity change signals into desensitized signals through the Brillouin scattering frequency shift effect; three-dimensionally encode the desensitized signals based on a Mach-Zehnder interferometer array to obtain a nonlinear mapping of the physical quantity change signals in the feature space; based on the nonlinear mapping of the feature space, a diffraction neural network is constructed using a preset phase-changing material; the physical quantity change signals are input into the diffraction neural network, and the diffraction neural network realizes iterative update of the weight matrix based on the back-propagation optical path.

[0016] The early warning module is used to decouple the four-dimensional sensing quantity in the physical quantity change signal through a diffraction neural network, and analyze and output the early warning signal through a microring resonant cavity array.

[0017] In the third aspect of the present application, an electronic device is provided, including a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device performs any of the methods described above.

[0018] In a fourth aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and the computer program is executed by a processor to perform any of the above methods.

[0019] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0020] 1. The physical quantity change signal is converted into a desensitized signal through the Brillouin scattering frequency shift effect; the desensitized signal is three-dimensionally encoded to obtain the nonlinear mapping of the physical quantity change signal in the feature space; a diffraction neural network is constructed using a preset phase-changing material; the physical quantity change signal is input into the diffraction neural network, and the diffraction neural network implements iterative updates of the weight matrix based on the back-propagation optical path; the four-dimensional sensing quantity in the physical quantity change signal is decoupled through the diffraction neural network, and the warning signal is output through micro-ring resonant cavity array analysis. Through the deep coupling of optical physical mechanism and algorithm logic, an optical domain closed loop is achieved in the hardware deployment, dynamic learning and decision output links, thereby improving the fidelity of the sensing signal in complex electromagnetic environments, the real-time response of the system, and the accuracy of abnormal event judgment.

[0021] 2. Based on the Brillouin scattering radiofrequency shift effect of the Bragg grating array, the Brillouin scattering radiofrequency shift is calculated from the physical quantity change signal; based on the Brillouin scattering radiofrequency shift, the temperature-strain cross-sensitivity in the physical quantity change signal is eliminated through the double-pulse difference method to convert the physical quantity change signal into a desensitized signal, thereby eliminating its cross-sensitivity interference and providing a clear and identifiable input basis for subsequent three-dimensional interferometric coding.

[0022] 3. A joint distribution model of four-dimensional sensing quantities is established by constructing vibration equations, temperature-strain coupling equations, and chemical concentration kinetics equations. The joint distribution model is decoupled through tensor decomposition of the diffraction neural network, thereby extracting the independent response mode of each physical disturbance quantity in the high-dimensional coupling field, eliminating the interference terms and nonlinear cross-influences between multiple physical fields, improving the separability and identifiability of abnormal disturbance sources in complex sensing environments, and realizing high-precision independent perception of multiple physical disturbances such as vibration, temperature, strain, and chemical concentration. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a flow chart of an intelligent monitoring method based on optical fiber sensing and optical neural network provided in an embodiment of the present application;

[0024] Figure 2 This is a schematic diagram of the heterogeneous integration topology of a fiber optic sensor network and a photonic chip provided in an embodiment of the present application;

[0025] Figure 3 This is a schematic diagram of a tunable photonic crystal unit composed of a Ge2Sb2Te5 phase change material provided in an embodiment of the present application;

[0026] Figure 4 This is a module schematic diagram of an intelligent monitoring device based on optical fiber sensing and optical neural network provided in an embodiment of the present application;

[0027] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0028] Explanation of the reference numerals: 41, processing module; 42, early warning module; 501, processor; 502, communication bus; 503, user interface; 504, network interface; 505, memory. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments.

[0030] The terms used in the following examples of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification of the present application, the singular expressions "a", "an", "said", "above", "the", and "this" are intended to include plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in the present application refers to and includes any or all possible combinations of one or more of the listed items.

[0031] In the following, the terms "first" and "second" are used for descriptive purposes only and should not be understood to imply or suggest relative importance or implicitly indicate the number of the technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of this application, unless otherwise specified, "plurality" means two or more.

[0032] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0033] Please refer to Figure 1 , which shows a flow chart of an intelligent monitoring method based on optical fiber sensing and optical neural network provided in an embodiment of the present application, the flow chart mainly includes the following steps: S101 to S105.

[0034] Step S101 : collecting a physical quantity change signal based on a Bragg grating array, and converting the physical quantity change signal into a desensitized signal through a Brillouin scattering frequency shift effect.

[0035] Specifically, the Bragg grating array is used to collect physical quantity change signals, which are optical signals. The Brillouin scattering frequency shift effect is used to convert the signals into desensitized signals after decoupling the temperature and strain factors, providing high-fidelity input for subsequent optical domain feature encoding and intelligent reasoning. Among them, the physical quantity change signals include: strain (ε): which indicates the relative change in the shape or size of an object under the action of force, such as the slight elongation or compression of the pipe wall due to pressure or mechanical load in industrial pipelines; temperature (T): related to the thermal state of the object, such as in a high-temperature reactor, temperature changes will affect the rate of chemical reactions and product distribution; vibration (A): the reciprocating motion of an object near its equilibrium position, such as the vibration of mechanical rotating equipment, which can be used to monitor the operating status of the equipment and detect potential faults in advance; chemical concentration (C): the content of a specific chemical substance in a mixture, such as in chemical production, monitoring the concentration of a certain reactant or product in the reactor to ensure that the reaction progress is as expected.

[0036] In a possible embodiment, step S101 also includes: collecting a physical quantity change signal based on a Bragg grating array, and converting the physical quantity change signal into a desensitized signal through the Brillouin scattering frequency shift effect, specifically including: calculating the Brillouin scattering frequency shift amount through the physical quantity change signal based on the Brillouin scattering frequency shift effect of the Bragg grating array; based on the Brillouin scattering frequency shift amount, eliminating the temperature-strain cross sensitivity in the physical quantity change signal through a double-pulse differential method to convert the physical quantity change signal into a desensitized signal.

[0037] Specifically, based on the Brillouin scattering frequency shift effect of the Bragg grating array, a calculation relationship between the physical quantity change signal and the Brillouin scattering frequency shift is constructed:

[0038] ;

[0039] in, is the Brillouin scattering frequency shift, is the Brillouin coefficient , is the strain disturbance, is the thermo-optical coefficient, The temperature change is measured using a double-pulse differential method to eliminate temperature-strain cross-sensitivity, achieving an error suppression rate of 99.7%. Since temperature and strain act together on the frequency shift response, to eliminate their cross-sensitivity interference, a double-pulse differential method is further employed for directional excitation and to obtain the forward and reverse differential values ​​of the Brillouin response. This constructs a decoupling model of temperature and strain, thereby extracting a frequency shift signal with a single physical quantity response characteristic, known as the desensitized signal. This desensitized signal is output as an optical signal, providing a clearly structured and recognizable input foundation for subsequent 3D interferometric encoding.

[0040] Step S102 : performing three-dimensional encoding on the desensitized signal based on a Mach-Zehnder interferometer array to obtain a nonlinear mapping of the physical quantity change signal in a feature space.

[0041] Specifically, the desensitized signal is three-dimensionally encoded based on the Mach-Zehnder interferometer array. The three-dimensional encoding is time-frequency-phase three-dimensional encoding. The transmission matrix of the time-frequency-phase three-dimensional encoding realizes the interference reconstruction between different paths by adjusting the phase difference. Its expression is:

[0042] ;

[0043] in, represents a nonlinear mapping of the feature space, represents the phase difference, Is the imaginary unit, used in mathematics and physics to represent the imaginary part of a complex number. For desensitization signal, is the propagation constant, which indicates the rate at which the phase of a light wave changes with distance when it propagates in a medium. Its unit is radian / meter (rad / m). The optical path length represents the actual path length of light propagating in the medium. In optical fibers, the optical path length is related to the physical length and refractive index of the optical fiber and is used to calculate the phase accumulation and energy loss of light waves in the optical fiber.

[0044] Step S103: constructing a diffraction neural network based on nonlinear mapping of the feature space and using a preset phase-changing material.

[0045] Specifically, based on the nonlinear mapping of the feature space, it is guided as the input light field into the diffraction neural network constructed by the preset phase-changing material, the unitary transformation is performed using a multi-layer optical matrix, and the real-time iterative update of the weight matrix is ​​achieved through the optical domain back propagation mechanism, thereby completing the deep reasoning and adaptive learning of the signal.

[0046] In a possible embodiment, step S103 further includes: guiding the nonlinear mapping of the feature space to an optical matrix processor, the optical matrix processor being composed of a Mach-Zehnder interferometer array; performing a multi-layer unitary transformation operation based on a weight matrix through each layer of the optical matrix processor, and guiding the weight matrix to be iteratively updated based on the phase change generated in the back-propagation optical path, so as to construct a diffraction neural network using a preset phase-changing material, the preset phase-changing material being .

[0047] Specifically, the characteristic space nonlinear mapping light field generated by the Mach-Zehnder interferometer array is used as input and guided to the optical matrix processor, which consists of multiple diffraction layers. The internal structure of the diffraction layer is constructed based on the programmable Mach-Zehnder interferometer, and the unitary transformation of the complex matrix is ​​realized by controlling the phase difference; in the forward propagation process of the optical signal, the input light field is sequentially acted upon by multiple layers of unitary matrices to complete the nonlinear mapping of the light field intensity and phase structure; at the same time, the error signal is introduced through the reverse propagation optical path, and the phase conjugate light field generated by stimulated Brillouin scattering is used to dynamically control the weight kernel function in the diffraction layer. The weight is determined by the integrated optical waveguide structure. Phase change materials are used to induce changes in their refractive index through the two-photon absorption mechanism, thereby finely adjusting the transmission characteristics of each layer of diffraction patterns, realizing the gradual iterative optimization of the weight matrix in the optical domain, and completing the learning and modeling of complex perturbation patterns. Among them, each layer of optical matrix processor performs multi-layer unitary transformation operations based on the weight matrix through the following formula. Each layer of the optical matrix processor receives the input light field distribution of the nth layer and processes the input light field distribution under the action of a nonlinear function; through the integral mapping relationship with the weight matrix of the nth layer, it generates the output light field distribution of the n+1th layer:

[0048]

[0049] in, Indicates the Output light field distribution of the layer optical matrix processor, Indicates the The input light field distribution of the layer optical matrix processor, For the The weight matrix corresponding to the layer optical matrix processor, is a nonlinear function, represents the input light field, Represents the output light field.

[0050] In step S104, the physical quantity change signal is input into the diffraction neural network, and the diffraction neural network implements iterative update of the weight matrix based on the back-propagation optical path.

[0051] Specifically, the light field formed by nonlinear mapping is input into the diffraction neural network, and intelligent reasoning and processing of multi-source physical disturbances are completed in the optical domain. The weight matrix is ​​iteratively updated through the back-propagation optical path, and the multi-dimensional coupling information in the physical quantity change signal is decoupled and reconstructed by combining tensor decomposition technology. The weight matrix is ​​determined by the hologram etching depth:

[0052]

[0053] in, For the The weight matrix corresponding to the layer optical matrix processor, is the imaginary unit, is the wavelength of light, is the effective refractive index, is the phase change.

[0054] Step S105 , decoupling the four-dimensional sensing quantity in the physical quantity change signal through a diffraction neural network, and analyzing and outputting a warning signal through a micro-ring resonant cavity array.

[0055] Specifically, the four-dimensional sensing quantities (strain, temperature, vibration, and chemical concentration) contained in the physical quantity change signal are decoupled through a diffraction neural network, and the decoupled signal is analyzed and output based on a microring resonant cavity array to generate a multi-level warning signal.

[0056] In a possible embodiment, step S105 further includes: establishing a joint distribution model of four-dimensional sensing quantities by constructing a vibration equation, a temperature-strain coupling equation, and a chemical concentration kinetic equation; and decoupling the joint distribution model by tensor decomposition of a diffraction neural network.

[0057] Specifically, first, a joint distribution model of the four-dimensional sensor quantities is constructed based on a priori physical mechanisms. This includes the vibration equation to characterize the spatiotemporal propagation characteristics of the structural response, the temperature-strain coupling equation to describe the coupled response relationship between the Brillouin frequency shift and the thermal and stress fields, and the chemical concentration kinetic equation to model the dynamic impact of concentration changes in the external environment on the signal. The expression of the joint distribution model is as follows:

[0058] ;

[0059] Among them, for the vibration equation, Represents vibration displacement, which is position and time function, Indicates the spatial position, usually along the axial direction of the optical fiber or structure, Indicates time, represents the wave velocity, that is, the speed at which the vibration wave propagates in the medium. It represents the damping coefficient, which reflects the energy dissipation during the vibration process.

[0060] For the temperature-strain coupling equation, is the Brillouin frequency shift, which is used to represent the change in optical frequency caused by temperature or strain changes. is the Brillouin frequency shift coefficient caused by temperature, which is used to indicate the degree of influence of temperature change on the Brillouin frequency shift. is the temperature change, used to indicate the temperature change. is the Brillouin frequency shift coefficient caused by strain, which is used to indicate the influence of strain change on the Brillouin frequency shift. is the strain change, which is used to represent the change of strain.

[0061] For the chemical concentration kinetic equation, is the time constant, which is used to indicate the time required for the system to reach steady state. is the concentration of the chemical substance, is the time function, is the reaction rate constant, which indicates the degree of influence of chemical reaction on concentration change. It is an input item that represents the impact of external factors (such as light, temperature, etc.) on the concentration of chemical substances.

[0062] Decoupling the joint distribution model through tensor decomposition of diffractive neural networks:

[0063] ;

[0064] in, represents the decoupled output tensor, i.e., the independent representation of the four-dimensional sensory quantities (vibration, temperature-strain coupling, and chemical concentration) after processing by the optical neural network. represents the feature rank, i.e., the number of features retained in the tensor decomposition. The accuracy and complexity of decoupling are determined by the feature rank. = 6, the sensing accuracy can be maintained at 98.3%. It represents the weight or coefficient of the rth feature, reflecting the importance of the feature in the decoupling process. It represents the weight or coefficient of the r-th feature, reflecting the importance of the feature in the decoupling process. 、 、 、 Respectively represent The patterns or distributions of the features in four dimensions (vibration, temperature-strain coupling, chemical concentration) are extracted from the original physical quantity change signals through tensor decomposition. Represents the tensor product (outer product) operation, which is used to combine feature patterns in different dimensions into a high-dimensional tensor.

[0065] Through experiments, the technical solution in this application has the following advantages over traditional detection systems: latency is reduced by 2 orders of magnitude (<100μs); energy efficiency is improved by 87% (0.3pJ / operation); concurrent processing of 10,000+ sensor nodes is supported; and the environmental adaptability temperature range is extended to -200℃~900℃.

[0066] The overall steps of the solution are further described below with reference to the accompanying drawings.

[0067] Hardware deployment phase: Please refer to Figure 2 , which shows a topological diagram of heterogeneous integration of a fiber optic sensor network and a photonic chip provided by an embodiment of the present application. In the left half of the figure: a spirally wound microstructured fiber array, showing the relationship between the FBG spacing Λ and the scattered light wavelength The right half of the relationship: the internal structure of the photonic chip, including the Mach-Zehnder modulation area (green) and the diffraction neural network area (blue). In this process, a microstructured optical fiber array is first laid out in a spiral winding manner along the surface of the device to be monitored (such as the outer wall of a high-temperature reactor). A single optical fiber integrates 1024 FBG sensing units, and the spacing between adjacent gratings is Λ=5 mm (determined by the gradient resolution of the target physical field). The optical fiber adopts a double-cladding structure, with an inner core diameter of 8.2 μm and a numerical aperture of 0.15, which is used to transmit the sensing light signal; the outer layer is a germanium-doped stress-sensitive layer with a thickness of 2 μm, and a periodic refractive index modulation is formed by femtosecond laser writing ( The photonic chip utilizes a silicon-on-silicon nitride heterogeneous integration process, etching a 256-channel Mach-Zehnder interferometer array (waveguide cross-section 800 × 400 nm², transmission loss <0.1 dB / cm) and a neural network processor consisting of four layers of diffraction holograms onto a 3×3 mm² chip. The optical fiber array is connected to the photonic chip via an inverted tapered coupler (taper angle 0.5°, coupling efficiency >98%), forming an optical-to-optical fusion signal link.

[0068] System initialization phase: please refer to Figure 3 , which shows a method provided by the embodiment of the present application Schematic diagram of a tunable photonic crystal unit made of phase change material. In the figure, the surface layer is 300nm thick. Phase change layer, refractive index n can be switched between 2.8-4.3; middle layer: silicon dioxide optical waveguide, cross-sectional size 800nm×400nm; bottom layer: silicon nitride stress adjustment layer, used to compensate for thermal deformation. In this process, a 1550 nm optical frequency comb (bandwidth 40 nm, linewidth 100kHz) is injected to establish a reference light field distribution, and the micro-ring resonator ( ) to calibrate the initial phase of each sensing channel. In the training phase, a controlled light sequence with a pulse width of 10 ns and a peak power of 20 mW is used. Two-photon absorption effect of phase change materials (nonlinear coefficient ) induces a refractive index jump and realizes weight matrix initialization.

[0069] Online learning phase: During this process, the real-time sensing light pulse (repetition frequency 10 MHz) is split into two paths by a beam splitter: the main path (95%) enters the forward inference optical path, passes through the interference coding layer (generating the time-frequency-phase three-dimensional feature vector) and the four diffraction holographic layers in sequence. The auxiliary path (5%) enters the reverse propagation optical path synchronously through the delay line, and is stimulated by Brillouin scattering (pump power 15 dBm, gain coefficient = m / W) to generate a phase conjugate error signal. The gradient signal is used to control the refractive index of the phase change material in real time via an electro-optical modulator (40 GHz bandwidth). The weight update step size is η = 0.02λ / n (n is the number of iterations). After 20 iterations, the loss function converges to less than 3% of the initial value.

[0070] During the decision output phase, the output light intensity distribution is analyzed by a microring resonator array (radius 50 μm, free spectral range (FSR) = 1.6 nm) into 32-level warning signals. The microrings dynamically match the characteristic wavelength through the thermo-optical effect (tuning efficiency 1.2 nm / mW), achieving a wavelength demodulation accuracy of 0.1 pm. A synchronous output module integrates a 1×4 multimode interference coupler, aligning the raw optical data (10 Gbps) with the diagnostic report (JSON format) via an optical delay line (compensating for 2.3 ns of delay difference) before output, achieving a full optical domain decision-making closed loop. The system maintained a prediction accuracy of 97.6% and a signal-to-noise ratio (SNR) better than 48 dB even in a -180°C liquid nitrogen environment.

[0071] The present application adopts the above method to convert the physical quantity change signal into a desensitized signal through the Brillouin scattering frequency shift effect; three-dimensionally encode the desensitized signal to obtain a nonlinear mapping of the physical quantity change signal in the feature space; construct a diffraction neural network using a preset phase-changing material; input the physical quantity change signal into the diffraction neural network, and the diffraction neural network implements iterative update of the weight matrix based on the back-propagation optical path; decouple the four-dimensional sensing quantity in the physical quantity change signal through the diffraction neural network, and analyze and output the warning signal through the micro-ring resonant cavity array, thereby realizing an optical domain closed loop in the hardware deployment, dynamic learning and decision output links through the deep coupling of optical physical mechanism and algorithm logic, thereby improving the fidelity of the sensing signal in complex electromagnetic environment, the real-time response of the system and the accuracy of distinguishing abnormal events.

[0072] Please refer to Figure 4 , which shows a module schematic diagram of an intelligent monitoring device based on optical fiber sensing and optical neural network provided by an embodiment of the present application, the device includes a processing module 41 and an early warning module 42, wherein,

[0073] The processing module 41 is used to collect the physical quantity change signal based on the Bragg grating array, and convert the physical quantity change signal into a desensitized signal through the Brillouin scattering frequency shift effect; perform three-dimensional encoding on the desensitized signal based on the Mach-Zehnder interferometer array to obtain the nonlinear mapping of the physical quantity change signal in the feature space; construct a diffraction neural network based on the nonlinear mapping of the feature space and use a preset phase-changing material; input the physical quantity change signal into the diffraction neural network, and the diffraction neural network realizes the iterative update of the weight matrix based on the back-propagation optical path.

[0074] The early warning module 42 is used to decouple the four-dimensional sensing quantity in the physical quantity change signal through a diffraction neural network, and analyze and output the early warning signal through the micro-ring resonant cavity array.

[0075] In one possible embodiment, the processing module 41 is used to collect a physical quantity change signal based on a Bragg grating array, and convert the physical quantity change signal into a desensitized signal through the Brillouin scattering frequency shift effect. Specifically, it includes: calculating the Brillouin scattering frequency shift amount through the physical quantity change signal based on the Brillouin scattering frequency shift effect of the Bragg grating array; based on the Brillouin scattering frequency shift amount, eliminating the temperature-strain cross sensitivity in the physical quantity change signal through a double-pulse differential method to convert the physical quantity change signal into a desensitized signal.

[0076] In one possible embodiment, the processing module 41 is used to perform three-dimensional encoding on the desensitized signal based on the Mach-Zehnder interferometer array, specifically including: three-dimensional encoding on the desensitized signal based on the Mach-Zehnder interferometer array, the three-dimensional encoding is time-frequency-phase three-dimensional encoding, and the transmission matrix of the time-frequency-phase three-dimensional encoding realizes interference reconstruction between different paths by adjusting the phase difference.

[0077] In one possible embodiment, the processing module 41 is used to construct a diffraction neural network based on the nonlinear mapping of the feature space and using a preset phase-changing material, specifically including: guiding the nonlinear mapping of the feature space to an optical matrix processor, the optical matrix processor being composed of a Mach-Zehnder interferometer array; performing multi-layer unitary transformation operations based on a weight matrix through each layer of the optical matrix processor, and guiding the weight matrix to be iteratively updated based on the phase change generated in the back-propagation optical path, so as to construct a diffraction neural network using a preset phase-changing material.

[0078] In one possible embodiment, the processing module 41 is used to perform multi-layer unitary transformation operations based on the weight matrix through each layer of optical matrix processor, specifically including: each layer of optical matrix processor receives the input light field distribution of the nth layer, and processes the input light field distribution under the action of a nonlinear function; and generates the output light field distribution of the n+1th layer through an integral mapping relationship with the weight matrix of the nth layer.

[0079] In a possible implementation, the processing module 41 is used for the diffraction neural network to implement iterative updating of the weight matrix based on the back-propagation optical path, specifically including: the weight matrix is ​​determined by the hologram etching depth.

[0080] In one possible embodiment, the early warning module 42 is used to analyze the four-dimensional sensor quantities in the physical quantity change signal through a diffraction neural network, specifically including: establishing a joint distribution model of the four-dimensional sensor quantities by constructing a vibration equation, a temperature-strain coupling equation, and a chemical concentration kinetics equation; and decoupling the joint distribution model through tensor decomposition of the diffraction neural network.

[0081] It should be noted that the above embodiments provide devices that implement their functions using only the division of the above functional modules as examples. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiments are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0082] This application also provides an electronic device. Figure 5 , Figure 5 5 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include: at least one processor 501, at least one communication bus 502, a user interface 503, at least one network interface 504, and a memory 505.

[0083] The communication bus 502 is used to implement the connection and communication between these components.

[0084] The user interface 503 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 503 may also include a standard wired interface and a wireless interface.

[0085] The network interface 504 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0086] The processor 501 may include one or more processing cores. Using various interfaces and circuits, the processor 501 connects to various components within the server. It executes instructions, programs, code sets, or instruction sets stored in the memory 505, as well as accesses data stored in the memory 505, to perform various server functions and process data. Optionally, the processor 501 may be implemented using at least one of the following hardware forms: a digital signal processing (DSP), a field-programmable gate array (FPGA), or a programmable logic array (PLA). The processor 501 may integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing content displayed on the display screen; and the modem handles wireless communications. It is understood that the modem may also be implemented as a separate chip, rather than integrated into the processor 501.

[0087] Among them, the memory 505 may include a random access memory (RAM) or a read-only memory (Read-Only Memory). Optionally, the memory 505 includes a non-transitory computer-readable storage medium. The memory 505 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 505 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 505 may also be optionally at least one storage device located away from the aforementioned processor 501. Reference Figure 5 , as a computer storage medium, the memory 505 may include an operating system, a network communication module, a user interface module, and an intelligent monitoring application based on optical fiber sensing and optical neural network.

[0088] exist Figure 5In the electronic device shown, the user interface 503 is mainly used to provide an input interface for the user and obtain the data input by the user; and the processor 501 can be used to call the intelligent monitoring application based on optical fiber sensing and optical neural network stored in the memory 505. When executed by one or more processors 501, the electronic device executes one or more of the methods described in the above embodiments. It should be noted that for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should know that this application is not limited to the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for this application.

[0089] The present application also provides a computer-readable storage medium storing instructions, which, when executed by one or more processors, enable an electronic device to execute one or more of the methods described in the above embodiments.

[0090] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0091] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are merely schematic, such as the division of units, which is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interface, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0092] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0093] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0094] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of this application, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the method of this application. The aforementioned memory includes various media that can store program code, such as USB flash drives, mobile hard drives, magnetic disks, or optical disks.

[0095] The above descriptions are merely exemplary embodiments disclosed in this application and are not intended to limit the scope of this application. That is, any equivalent changes and modifications made based on the teachings disclosed in this application are still within the scope of this application.

[0096] This application is intended to cover any modifications, uses or adaptations disclosed in this application, which follow the general principles disclosed in this application and include common knowledge or customary technical means in the technical field not disclosed in this application.

Claims

1. An intelligent monitoring method based on optical fiber sensing and optical neural network, characterized in that: The method comprises: The physical quantity change signal is collected based on a Bragg grating array, and the physical quantity change signal is converted into a desensitized signal through the Brillouin scattering frequency shift effect; Performing three-dimensional encoding on the desensitized signal based on a Mach-Zehnder interferometer array to obtain a nonlinear mapping of the physical quantity change signal in a feature space; Based on the nonlinear mapping of the feature space, a diffraction neural network is constructed using a preset phase-changing material; Inputting the physical quantity change signal into the diffraction neural network, and the diffraction neural network implements iterative update of the weight matrix based on the back-propagation optical path; Decoupling the four-dimensional sensing quantity in the physical quantity change signal through the diffraction neural network, and analyzing and outputting the warning signal through the micro-ring resonant cavity array; The method collects a physical quantity change signal based on a Bragg grating array and converts the physical quantity change signal into a desensitized signal through the Brillouin scattering frequency shift effect, specifically comprising: calculating a Brillouin scattering frequency shift amount from the physical quantity change signal based on the Brillouin scattering frequency shift effect of the Bragg grating array; eliminating temperature-strain cross-sensitivity in the physical quantity change signal through a double-pulse difference method based on the Brillouin scattering frequency shift amount, so as to convert the physical quantity change signal into the desensitized signal; and decoupling a joint distribution model through tensor decomposition of the diffraction neural network: ； in, represents the decoupled output tensor, 、 、 、 Respectively represent The pattern or distribution of features in four dimensions: vibration, temperature, strain coupling, and chemical concentration. represents the tensor product operation, is the feature rank.

2. The method according to claim 1, characterized in that The three-dimensional encoding of the desensitized signal based on the Mach-Zehnder interferometer array specifically includes: The desensitized signal is three-dimensionally encoded based on the Mach-Zehnder interferometer array, wherein the three-dimensional encoding is time-frequency-phase three-dimensional encoding, and the transmission matrix of the time-frequency-phase three-dimensional encoding realizes interference reconstruction between different paths by adjusting the phase difference.

3. The method according to claim 1, characterized in that The nonlinear mapping based on the feature space and the use of a preset phase-changing material to construct a diffraction neural network specifically include: The nonlinear mapping of the feature space is used as a guide to an optical matrix processor, wherein the optical matrix processor is composed of the Mach-Zehnder interferometer array; The optical matrix processor of each layer performs multi-layer unitary transformation operations based on the weight matrix, and guides the weight matrix to be iteratively updated based on the phase change generated in the back-propagation optical path, so as to construct the diffraction neural network using a preset phase-changing material.

4. The method according to claim 3, characterized in that The performing of a multi-layer unitary transformation operation based on the weight matrix by the optical matrix processor at each layer specifically includes: The optical matrix processor of each layer receives the input light field distribution of the nth layer and processes the input light field distribution under the action of a nonlinear function; The output light field distribution of the n+1th layer is generated through the integral mapping relationship with the weight matrix of the nth layer.

5. The method according to claim 3, characterized in that The diffractive neural network implements iterative updating of the weight matrix based on the back-propagation optical path, specifically including: The weight matrix is ​​determined by the hologram etching depth.

6. The method according to claim 1, characterized in that The four-dimensional sensing quantity in the physical quantity change signal by the diffraction neural network specifically includes: The joint distribution model of the four-dimensional sensing quantity is established by constructing a vibration equation, a temperature-strain coupling equation and a chemical concentration kinetic equation; The joint distribution model is decoupled by tensor decomposition of the diffractive neural network.

7. An intelligent monitoring device based on optical fiber sensing and optical neural network, characterized in that: The device includes a processing module and an early warning module, wherein: The processing module is used to collect physical quantity change signals based on a Bragg grating array, and convert the physical quantity change signals into desensitized signals through the Brillouin scattering frequency shift effect; perform three-dimensional encoding on the desensitized signal based on a Mach-Zehnder interferometer array to obtain a nonlinear mapping of the physical quantity change signal in a feature space; construct a diffraction neural network based on the nonlinear mapping of the feature space and using a preset phase-changing material; input the physical quantity change signal into the diffraction neural network, and the diffraction neural network implements iterative update of the weight matrix based on a back-propagation optical path; the physical quantity change signal is collected based on the Bragg grating array, and the physical quantity change signal is converted into a desensitized signal through the Brillouin scattering frequency shift effect, specifically including: calculating the Brillouin scattering frequency shift amount based on the physical quantity change signal based on the Brillouin scattering frequency shift effect of the Bragg grating array; and eliminating the temperature-strain cross sensitivity in the physical quantity change signal through a double-pulse difference method according to the Brillouin scattering frequency shift amount, so as to convert the physical quantity change signal into the desensitized signal; The warning module is used to decouple the four-dimensional sensor quantity in the physical quantity change signal through the diffraction neural network, and output the warning signal through micro-ring resonator array analysis; and decouple the joint distribution model through tensor decomposition of the diffraction neural network: ； in, represents the decoupled output tensor, 、 、 、 Respectively represent The pattern or distribution of features in four dimensions: vibration, temperature, strain coupling, and chemical concentration. represents the tensor product operation, is the feature rank.

8. An electronic device, characterized in that: The electronic device comprises a processor, a communication bus, a user interface, a network interface and a memory, wherein the memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores instructions, and when the instructions are executed, the method according to any one of claims 1 to 6 is performed.

Images