Multi-structure light feedback detection device for fiber-optic illumination system

Abstract

The application discloses a multi-structure light feedback detection device of an optical fiber lighting system and belongs to the technical field of optical fiber sensing. A polarization-independent bidirectional light path is constructed by using an integrating sphere combining wave cavities, a mode field collimating polarization matching channel and a wide-band multi-mode optical circulator to realize high-isolation reverse light extraction. A diffraction-preventing overlapping boundary is derived, and a micro-slit array is designed to eliminate band crosstalk. A wavelength and space joint feature matrix is generated by using bottom-layer FPGA clock synchronization multi-source heterogeneous sampling to realize working condition attribution. In combination with a stable boundary constraint differential proportional integral control law, a temperature-microbend cross-sensitivity decoupling model and a total variation attenuation distribution reconstruction engine, damaged wavelengths are accurately and independently compensated. The application effectively eliminates light path crosstalk and temperature drift misjudgment, does not increase nodes to realize sub-decimeter fault positioning and early warning, and improves environmental adaptability and detection precision.

Description

Technical Field

[0001] This invention belongs to the field of fiber optic sensing technology, specifically a multi-structured light feedback detection device for fiber optic lighting systems. Background Technology

[0002] Fiber optic lighting technology, as an important branch of the lighting field, has developed rapidly in recent years. Due to its advantages such as high-efficiency transmission, flexible deployment, energy saving, and environmental friendliness, fiber optic lighting systems have been widely used in diverse scenarios such as stage lighting, automotive lighting, medical lighting, and architectural decoration. With the continuous enrichment and complexity of application scenarios, higher requirements are being placed on the operational status monitoring and fault diagnosis of fiber optic lighting systems.

[0003] In fiber optic lighting systems, optical feedback detection technology is a crucial element for system status monitoring. The optical feedback signal carries important information such as fiber optic transmission characteristics, light source operating status, and optical path coupling efficiency. By analyzing and processing this information, the health status of the lighting system can be assessed and early warnings issued. However, existing optical feedback detection schemes still have significant shortcomings in terms of detection dimensions and intelligence, making it difficult to meet the diverse detection needs of complex application scenarios.

[0004] A search revealed a polarization feedback system disclosed in patent document CN110542025A, published on March 29, 2024. This system includes a laser source, an integrated lens, an optical fiber, a light source module, a detector, and a control device. The integrated lens consists of a coupling lens and a feedback lens located outside the coupling lens. This technical solution collects the reverse-transmitted light from the optical fiber through the feedback lens and guides it to the detector for optical signal analysis. However, the feedback structure in this solution is relatively simple, only capable of polarization feedback. The detector can only acquire polarized light energy or wavelength information, failing to meet the requirement of simultaneous detection of multiple optical feedback parameters. Furthermore, the control device of this system has limited functionality, only controlling the on / off state of the laser source, without comprehensive control over the overall operating state of the lighting system, resulting in a low level of intelligence.

[0005] A search revealed a fiber optic feedback system disclosed in patent document CN110701503A, published on January 17, 2020. This system includes a laser coupling device, a dichroic mirror, an optical fiber, a light source module, a converging lens, a detector, and a control device. This technical solution employs an in-fiber feedback method, using a dichroic mirror to split white light and guide the reflected light to the detector to acquire the optical feedback signal. This integrated packaging design offers the advantage of a compact structure. However, the acquisition of optical feedback in this solution relies on the splitting characteristics of the dichroic mirror, and the feedback path and structure are relatively fixed, making flexible configuration of multi-structure optical feedback impossible. Furthermore, the detector can only detect reflected light energy or single wavelength information, limiting the dimension of feedback information and failing to provide comprehensive decision-making basis for the optimized control of the lighting system.

[0006] Based on the above-mentioned existing technical solutions, it can be seen that existing fiber optic lighting systems mainly suffer from the following technical problems in optical feedback detection: First, the feedback structure is simple, mostly using fixed beam splitters or polarizers, making it difficult to achieve flexible configuration of multi-structured optical feedback; second, the detection dimensions are limited, usually only acquiring light energy or single wavelength information, unable to obtain rich state information from the spectral dimension; third, the level of intelligence is insufficient, the control logic is simple, and there is a lack of multi-source data fusion and comprehensive judgment mechanisms, making it impossible to achieve a comprehensive and accurate assessment of the system's health status. These technical problems limit the reliability and maintainability of fiber optic lighting systems in complex application scenarios, making it difficult to meet the increasingly diverse detection needs.

[0007] Therefore, this paper proposes a multi-structured optical feedback detection system for fiber optic lighting systems. Through innovative multi-wavelength channel spectral splitting and independent detection architecture, combined with a multi-source data fusion-based fault comprehensive judgment method, it achieves comprehensive, real-time, accurate detection and intelligent monitoring of the operating status of fiber optic lighting systems, which has significant technical significance and application value. Summary of the Invention

[0008] The purpose of this invention is to provide a multi-structured light feedback detection device for fiber optic lighting systems, in order to solve the problems of existing technologies mentioned in the background art, such as single feedback structure, limited detection dimensions, and inability to achieve accurate evaluation.

[0009] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A multi-structured optical feedback detection device for a fiber optic lighting system, comprising: A multi-channel assembled light source module, comprising multiple independently driven light-emitting chips with different center wavelengths; The integrating sphere combining cavity receives the light beam emitted by the light-emitting chip at its inlet and completes multi-wavelength spatial continuous combining in a polarization-independent manner. A multimode self-collimating optical circulator has a first port, a second port, and a third port. The first port is connected to the output port of the integrating sphere multiplexer, and the second port is connected to the main monitoring link multimode fiber that carries multiple discrete spatial sensing nodes. The optical circulator is configured to transmit the forward multiplexed beam from the first port to the second port with low loss, and to directionally route the random polarization feedback beam returned by the self-collimating link fiber to the third port to achieve directional isolation of the first port. A precision anti-diffraction spatial beam splitting module, connected to the third port, is used to disperse the emitted reverse feedback light and output the optical signals of each wavelength channel through a micro-slit array. The data processing and control motherboard integrates a global real-time hardware clock triggering unit, which is configured to synchronously control the sampling actions of discrete spatial sensing nodes and the spectral sampling of the beam splitting module based on a reference triggering clock, achieve timing alignment, generate a wavelength and spatial joint feature matrix, and perform fault location, optical power closed-loop feedback compensation and early warning based on the matrix.

[0010] According to the above technical solution, a mode field collimation polarization matching channel is set between the output port of the integrating sphere multiplexing cavity and the first port, and between the second port and the main monitoring link fiber. The channel includes an aspherical aberration-correcting microlens array and a phase-shifting superposition region composed of a spatial birefringent crystal, a non-reciprocal magneto-optical rotator, and a half-wave plate. It is configured to compress, collimate, and decompose the forward diverging multimode beam into orthogonal polarization components, which are then superimposed by polarization and output to the second port in geometric space. It is also configured to cause the randomly polarized feedback light to undergo polarization state exchange and reverse geometric divergence when passing through in the reverse direction. The lateral offset spatial distance of the reverse geometric divergence is greater than the waveguide receiving aperture of the first port to suppress the backflow of feedback light into the light source side.

[0011] According to the above technical solution, the precision anti-diffraction spatial beam splitting module sequentially includes a collimating lens group, a planar scribed diffraction grating, an imaging focusing lens group, and a metal microstructure beam splitting substrate located at the focal plane. Independent microslits corresponding to the first-order diffraction deflection positions of each emission channel wavelength are formed on the metal microstructure beam-splitting substrate. The physical geometric center distance between two adjacent microslits on the dispersion axis is greater than the sum of the following two physical distances: the first is the sum of the actual geometric opening half-widths of the two adjacent independent microslits, and the second is the diffusion superposition threshold of the principal maximum Airy disk radius determined by the inner diameter of the focusing lens and the two adjacent wavelengths.

[0012] According to the above technical solution, the data processing and control motherboard broadcasts and triggers the distributed sensing nodes to instantaneously latch the light intensity sampling value through the underlying differential bus; at the same time, it is equipped with a high-speed analog-to-digital converter and a dual-port ring first-in-first-out queue, and performs a rectangular window sliding average downsampling operation at the moment of reaching the reference trigger to eliminate static dark current noise and generate a downsampling normalized spectral column vector aligned with the current time window. The system converts the light intensity scalar transmitted from the spatial sensing node into a dimensionless real-time transmission transmittance column vector by subtracting the factory-calibrated fixed background loss; and constructs a wavelength-spatial joint feature matrix by performing an outer product operation on the normalized spectral column vector and the transmittance column vector.

[0013] According to the above technical solution, the data processing and control motherboard is configured to expand the wavelength and spatial joint feature matrix into a one-dimensional real-time feature column vector in column priority order, and perform feature matching operation based on Gaussian statistical pattern recognition. Specifically, the deviation of the real-time feature column vector from the expected column vector of various typical working conditions in the standard feature database is calculated, and the weighted quadratic form mapping and square root are performed using the inverse matrix of the positive definite sample covariance matrix of the corresponding pattern to obtain the statistical Mahalanobis distance scalar.

[0014] According to the above technical solution, when the microbending loss state is identified, the data processing and control motherboard locates the abnormal broken node by calculating the maximum physical gradient difference of the transmittance of the spatial sensing node; extracts the whole column of light intensity loss slice vector at the node from the wavelength and spatial joint feature matrix, and calculates the dimensionless relative energy deficit ratio of each emission wavelength relative to the expected value of no loss. The system independently adjusts the driving current of each light-emitting chip by executing a discretized differential anti-saturation proportional-integral control law based on the energy deficit ratio. The proportional and integral control gain of the control law is controlled by the decoupling control stability boundary condition. The upper limit of this boundary is jointly determined by the light-emitting channel reference luminous energy, the differential slope of the electro-optic conversion efficiency, and the transmission attenuation coupling coefficient at the break point.

[0015] According to the above technical solution, the system selects a long wavelength channel to construct a continuously evolving dynamic state column vector, which includes the instantaneous equivalent logarithmic decay coefficient, first-order deterioration rate and second-order aging strain acceleration at the positioning node. The system is configured with a time-varying adaptive Kalman recursive model, based on the prior estimation of uniform acceleration state transition, and uses online measurement information and smooth forgetting factor to dynamically update the measurement noise variance, extract the converged second-order acceleration estimate, and use the continuous parabolic trajectory extrapolation formula to calculate the estimated logarithmic decay scalar after a set time delay in the future.

[0016] Based on the above technical solution, a two-parameter cross-sensitivity separation model is established. The instantaneous logarithmic macroscopic attenuation change of the short-wavelength and long-wavelength channels at the positioning node is extracted, and a sensitivity coupling matrix is ​​constructed using the pre-calibrated thermo-optical attenuation sensitivity coefficient and deformation attenuation sensitivity coefficient. By performing inverse matrix operations, the mixed attenuation variables are decoupled and separated into pure physical micro-bending rate variables and pure ambient temperature drift. The system only converts the decoupled pure physical micro-bending rate variable into an energy deficit ratio input to the proportional-integral control law for dynamic compensation.

[0017] Based on the above technical solution, the system has a pre-defined arbitration superposition rule for temperature compensation and dynamic control: When the decoupled ambient temperature drift exceeds the set background thermal compensation threshold, the system directly adjusts the static reference bias current of each wavelength constant current drive circuit according to the temperature lookup table; the proportional-integral control law only performs micro-perturbation dynamic superposition control for pure physical micro-bends on the adjusted static reference bias current.

[0018] According to the above technical solution, the system is equipped with a super-resolution positioning engine that reconstructs continuous spatial attenuation distribution. The system constructs an objective functional, whose independent variable is a piecewise smooth one-dimensional spatial decay distribution density function. The objective functional includes a data fidelity error term constructed from measured cumulative macroscopic loss values, and a total variational regularization sparsity penalty term composed of first-order spatial derivatives. The system solves the continuous decay density curve that minimizes the functional by finite element discretization partial differential iteration, and scans the maximum value of the spatial finite difference step of the curve to determine the corresponding spatial physical unit coordinates as the continuous physical location coordinates of the fault point.

[0019] Compared with the prior art, the present invention has the following beneficial effects: In this invention, a mode field collimation polarization matching channel based on an integrating sphere combined cavity, aspherical aberration-correcting collimation, and phase shift superposition of a birefringent crystal, combined with a broadband multimode optical circulator, achieves high-isolation reverse light extraction with extremely low forward loss, completely suppressing feedback light disturbances and eliminating the risk of resonance instability. A micro-slit array designed based on anti-diffraction overlap physical boundaries, combined with a high-pass dielectric film, absolutely isolates high-order diffraction energy at the physical level, eliminating band crosstalk and ensuring high purity of spectral acquisition. Utilizing a global hardware clock synchronization trigger on the underlying FPGA, high-speed spectral and low-frequency spatial sensing data are aligned within a unified time window, constructing a wavelength-spatial joint feature matrix. Combined with pattern recognition and Mahalanobis distance calculation, spatiotemporal misalignment is eliminated, achieving high-confidence fault condition attribution. It pioneered a limited differential decoupling control, which locks the physical breakpoint through spatial gradient difference and independently compensates only for the deficient band. The control gain is constrained by the Lyapunov stability boundary, suppressing overshoot and oscillation. At the same time, it uses the differentiated sensitivity of multiple wavelengths to thermo-optic effects and micro-bending radiation leakage to construct a full-rank coupling matrix, accurately separating temperature drift and physical micro-bending rate, avoiding miscompensation and overload burnout of undamaged wavelengths. Attached Figure Description

[0020] Figure 1 This is a feedback flowchart of the optical feedback device of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Example 1 In a specific implementation of a fully closed-loop optical path bidirectional isolation and precision spatial beam splitting integration, this device constructs a complete bidirectional isolation structure for the combined light source and the coupled optical path, addressing the technical problem of the inability to extract the reverse feedback light and the instability of the light source resonance caused by the shared physical port of the light-emitting end and the spectral analysis end in the existing fiber optic sensing scheme.

[0023] Specifically, the core components of the main optical path front end of this device include a multi-channel assembled light source module, an integrating sphere combining cavity, and a three-port wideband multimode autocollimating optical circulator. The multi-channel assembled light source module integrates three light-emitting chips driven by independent constant current driving chips, namely a blue light chip with a center wavelength of 450nm, a green light chip with a center wavelength of 550nm, and a red light chip with a center wavelength of 650nm.

[0024] Specifically, such as Figure 1As shown, the multi-wavelength beams emitted by the three independent light-emitting chips are simultaneously incident into an integrating sphere cavity with an inner wall coated with a high-reflectivity barium sulfate coating and an inner diameter of 25 mm, achieving multi-wavelength spatial continuous combining in a polarization-independent manner. Preferably, to solve the numerical aperture problem... To address the issues of polarization depolarization and isolation degradation in high-order oblique mode fields caused by the maximum divergence half-angle of large-core plastic multimode fiber with a diameter of 0.5 mm reaching 30° under broadband multimode transmission, a mode field collimation polarization matching channel, consisting of an aspherical aberration-correcting microlens array and a spatial birefringent crystal combination, is constructed between the output port of the integrating sphere combiner cavity and the first port of the optical circulator, as well as between the second port and the main link fiber.

[0025] Furthermore, the multimode diverging beam emitted from the combined beam passes through an aspherical aberration-correcting microlens array, compressing and limiting its maximum meridional divergence angle to within a 1.5° collimation boundary; subsequently, the collimated beam enters the multimode self-collimating optical circulator. Specifically, the following is the equivalent polarization transformation Jones matrix for forward and reverse propagation, which integrates the beam-splitting coordinate transformation effect of the birefringent crystal and only characterizes the overall transformation law of the optical field polarization state: The collimated multimode unpolarized natural light field emitted from the integrating cavity is characterized as two orthogonal components. First, the light passes through a first-order spatial birefringent crystal (with the optical axis perpendicular), splitting the natural light into mutually perpendicular ordinary (o-ray) and extraordinary (e-ray) spatially parallel components. Then, the light field passes through a phase-shift superposition region formed by a Faraday rotator and a half-wave plate. The Faraday rotator generates a 45° non-reciprocal magneto-optical polarization rotation, while the half-wave plate generates a -45° reciprocal polarization phase shift. During forward propagation, the two phase shifts cancel each other out (combined phase shift). Its forward transmission Jones matrix is: The polarization axis of the light field does not deflect. When passing through the second-order complementary birefringent crystal, the o-ray and e-ray undergo reverse geometric refraction shifts equal to those of the first order, resulting in complete geometric overlap of the two forward-propagating polarization components in space. This allows for lossless insertion and output to the second port, coupling into a 20m long, 1mm core diameter, numerical aperture crystal. In the main monitoring link plastic fiber with a loss of 0.5, the overall insertion loss of this forward main link path is controlled within... Within.

[0026] Furthermore, when the main link fiber transmits a feedback optical signal to the left in the reverse direction, it includes backscattered Rayleigh light, Fresnel reflected light, and higher-order guided modes of fault leakage. When the light field passes backward through the second-order birefringent crystal from the second port, it is decomposed into orthogonally polarized beams. When it passes backward through the phase shift region of the half-wave plate and the Faraday rotator, the non-reciprocal rotation direction of the Faraday rotator remains unchanged at +45°, while the reciprocal phase shift of the half-wave plate is superimposed in reverse to +45°. The polarization superposition of the two induces an orthogonal rotation of the polarization planes at +90°. The reverse transmission Jones matrix is ​​rigorously characterized as follows: This matrix shows that the polarization states of the ordinary and extraordinary rays returning in reverse are interchanged (i.e., o-rays become e-rays, and e-rays become o-rays). When the beam enters the first-order spatial birefringent wedge crystal in reverse, the original refractive drift cannot be canceled out and instead forms a double reverse geometric divergence drift. Specifically, the spatial separation distance of the lateral offset of the returning rays... Strictly determined by the following derivation: In the formula, The equivalent transmission refractive index of a birefringent wedge-shaped crystal; The preferred physical length of the effective optical axis of the internal optical rotation separation cavity is 18.5 mm. The geometric crystal wedge angle is configured as 12° (i.e. This causes the cumulative net lateral spatial offset generated by the reverse return beam to reach... The deflection displacement far exceeds the upper limit of the 1.0mm fiber waveguide aperture of the first port, so that any randomly polarized feedback light is uniquely routed to the third port. The measured increase in the forward and reverse polarization extinction ratio results in a directional isolation of more than 35dB, effectively preventing the feedback light from disturbing the light-emitting cavity. At the same time, the third port is sealed and connected to the input end of the spectral splitting module, completing the bidirectional closed loop of the single main link.

[0027] In a specific embodiment of precision diffraction-free spatial beam splitting, the back-feedback light, after exiting from the third port of the optical circulator, is parallelized by a double-separated achromatic collimating lens group with a focal length of 30mm and perpendicularly illuminates a plane-defined diffraction grating, wherein the grating constant is 833.3nm and the blaze wavelength is set to 550nm. Further, the first-order diffracted light emitted by dispersion is focused at the focal plane by a three-element Cook imaging focusing lens group with an effective aperture of 25.4mm and an equivalent focal length of 60mm. Specifically, based on the grating dispersion equation, the first-order deflection angles for each center wavelength correspond to 32.68° for blue light, 41.30° for green light, and 51.26° for red light. The center coordinates of the dispersion axis of the lens group's focal plane specifically correspond to 38.49mm for the blue light channel, 52.71mm for the green light channel, and 74.88mm for the red light channel. The minimum center-to-center distance occurs between the blue and green channels, with an absolute geometric distance of 14.22mm. Preferably, to ensure that the full width at half maximum (FWHM) of each spectral acquisition channel is strictly set at 25 nm, the physical geometric aperture widths of the microslits placed on the chromium substrate are limited to 2.14 mm for blue light, 2.40 mm for green light, and 2.88 mm for red light. Furthermore, based on the geometric half-width non-overlap criterion and the Airy spot intensity superposition law, this invention derives the anti-crosstalk boundary conditions, and the system's beam-splitting substrate satisfies the following inequality: In the formula, k is the channel number index and ; and Characterizes the coordinates of the center of the dispersion axis (mm); and Characterizes the actual geometric opening width of the microslit (mm); and Characterizes the center wavelength of the chip, which is converted to mm during calculation (i.e., multiplied by 10). -6 ); The equivalent focal length (mm) for focusing; The effective light-transmitting inner diameter (mm) is used. Specifically, substituting the closest neighbor parameters of the blue-green channels for calibration, the sum of the half-widths is 2.270 mm, and the theoretical diffraction superposition term is 1.22 × (450 + 550) × 10. -6 ×60 / 25.4≈0.00288mm, the limit boundary constraint is 2.273mm; the actual system spacing of 14.22mm far exceeds this threshold, proving that the system has clear isolation in geometric space. Combined with the high-pass dielectric film with an out-of-band cutoff depth of OD4 vacuum-deposited on the slit surface, the photodiode array can effectively resist high-order diffraction crosstalk in adjacent bands.

[0028] Example 2 In a specific implementation of multi-source, multi-clock hardware synchronization and feature statistical spatial processing, to address the time scale mismatch between the 50kHz sampling rate of the spectral acquisition front-end and the 10Hz refresh rate of the distributed sensing link, a field-programmable gate array (FPGA) chip with a global real-time hardware clock counter integrated on the motherboard is used to generate a timing reference trigger clock with a fixed period of 100ms. Specifically, at the instant of each discrete reference trigger, the FPGA broadcasts the trigger via the differential bus to each spatial sensing node on the distributed link to latch the light intensity sampling value and send back a data frame encapsulated with a timestamp; in parallel, three independent high-speed analog-to-digital converters... The single-step sampling time interval is continuously converted from analog to digital, and the spectral data corresponding to a 100ms window is buffered in a dual-port circular first-in-first-out queue with a depth of 5000 words inside the FPGA. Furthermore, at the moment of triggering, the FPGA internal hardware automatically performs a rectangular window moving average operation, subtracting the calibrated static dark current noise from the original digital scalar within the moving window, and further dividing by the difference between the limiting light intensity and the noise floor under the calibration of the light-emitting chip without external force loss, to calculate in real time the normalized spectral column vector that is strictly aligned with the current time window.

[0029] Furthermore, for the digital transmission scalar returned by the eight spatial sensing nodes of the reference time link, after deducting the noise floor, it is normalized and mapped relative to the total internal reflection saturation reference scalar to convert it into real-time transmission transmittance in the interval (0, 1.0]. Preferably, considering the background absorption and flange insertion loss along the waveguide axis, the system establishes a background attenuation propagation integral equation, using the background coefficient integral along the length and the joint loss as constants to eliminate the fixed loss in the stress-free state, so that the lossless normal reference of the fifth spatial sensing point of the deployment link strictly converges to 0.6500, ensuring the physical self-consistency of the fault residual. Specifically, the system will time The normalized spectral column vector and transmittance column vector, aligned between the two sides, are operated on using the outer product rule to construct a 3×8 wavelength and spatial joint feature matrix characterizing the global wavelength and spatial loss distribution. This matrix is ​​then expanded into a one-dimensional 24-dimensional real-time feature column vector in column-major order. Subsequently, this feature column vector is fed into a digital signal processor to perform a matching operation based on Gaussian statistical pattern recognition. The deviation of the real-time features from the expected vectors of various typical operating conditions is calculated. The statistical Mahalanobis distance scalar is obtained by taking the weighted quadratic root of the inverse covariance matrix of 24×24 positive definite samples under each mode to achieve high-confidence operating condition classification and attribution of environmental disturbances.

[0030] Example 3 In a decoupled differential closed-loop current regulation implementation based on spatial fault location, when the pattern recognition system determines that the current state belongs to waveguide microbending loss, the control core calculates the step difference of the transmittance vector between adjacent points and identifies and locates the segment corresponding to the maximum step difference as an abnormal breakpoint node. Further, the control unit directly extracts the entire three-channel slice vector at this node from the aforementioned feature matrix and calculates the relative energy deficit ratio of each wavelength compared to the expected lossless value. Specifically, to avoid integral saturation during continuous current regulation of the independent drive circuit, the constant current chip performs discretized differential anti-saturation proportional-integral control. The target current for the next sampling step is synthesized by adding the proportional differential term and the integral accumulation term to the actual operating current of the current cycle, and is sent to each drive chip every 0.1s. Preferably, to ensure that local decoupled current regulation does not experience overshoot oscillation under sudden microbending, the proportional and integral control gains strictly follow the multi-channel decoupled control stability boundary equation: In the formula, For discrete proportional adjustment of gain (mA); For discrete integral adjustment of gain (mA / s); The sampling step size is set to discrete control (0.1s). The reference constant-state luminous energy (mW) is calibrated for the kth luminous channel. The slope of the electro-optical conversion efficiency (mW / mA) calibrated by linear regression at the junction temperature of the light-emitting chip. This is the ratio of the measured transmitted energy at the point of failure to the initial energy (dimensionless transmission attenuation coefficient). The closed-loop control engine can implement precise quantitative flow control for the actual loss band while ensuring stability.

[0031] Furthermore, this device selects the long-wavelength red light channel, which is most sensitive to waveguide perturbation leakage, and calculates the local equivalent macroscopic logarithmic attenuation coefficient by combining the product of the total waveguide length and the return propagation of the positioning node from the light outlet. The instantaneous attenuation coefficient, velocity, and acceleration are then jointly constructed into a three-dimensional dynamic state column vector. Specifically, the system initiates a time-varying Sage-Husa adaptive Kalman recursive model. Based on the prior estimation of the uniformly accelerated state transition, it recursively updates the external time-varying measurement noise variance based on measurement innovation and smoothing forgetting factors, filtering random background noise to achieve a stable second-order acceleration estimate. Preferably, the system calls the parabolic trajectory extrapolation formula, continuously recursively calculating the final estimated attenuation scalar after 288,000 sampling delays (8 hours ahead) with a step size of 0.1s. This calculation is then used to immediately trigger an acoustic and optical hardware early warning upon exceeding the safety stretching critical threshold, achieving advanced protection against the risk of hidden fractures in long-distance blind zones.

[0032] Example 4 In a specific implementation method for the normalization construction of a feature database and closed-loop verification under harsh operating conditions, the ambient temperature is maintained at... Relative humidity controlled at One hundred physical link systems with the same architecture as described above were built in the standard metrology laboratory. Specifically, the links were freely suspended with zero external force, and precision fixtures applied amplitude at different node sections. The serrated, slightly bent state, and the application of standard loop mold Under the macro-bending state of the critical radius of curvature, feature vectors are acquired at a frequency of 10Hz and expanded column-wise for each group of 10,000 frames. The empirical expected column vectors of each typical mode distribution and the 24×24 positive definite sample covariance matrix are then calculated. Preferably, during the calculation, the constant value is superimposed on the diagonal. The small regularization perturbation term prevents singular degradation in high-order modulus-related physical scenarios and ensures the numerical stability of the inverse operation of Mahalanobis distance solution.

[0033] Specifically, when the 5th sensor point experienced a micro-bending disturbance during retesting, the processing unit aligned and calculated the blue, green, and red spectral channel data as [0.82, 0.96, 0.78]. T The transmittance of the sensor nodes was simultaneously calculated to be [0.99, 0.98, 0.98, 0.97, 0.65, 0.64, 0.63, 0.62]. T The scan revealed that the largest difference jump occurred between points 4 and 5 (difference 0.32), identifying the abnormal breakpoint index as 5; the true loss slice of node 5 was directly extracted from the feature matrix [0.5330, 0.6240, 0.5070]. T Compared to the lossless expected baseline [0.6500, 0.6500, 0.6500] T The relative energy deficit ratios were found to be 18.0% for blue light, 4.0% for green light, and 22.0% for red light. The controller executes a discrete differential anti-saturation control law, issuing a decoupling command after three time steps (0.3s) to maintain a constant green light loop. The command instructs the constant current drive chip to add +27.0mA to the blue light chip and +33.0mA to the red light chip, pushing the global real-time characteristics and the lossless normal center Mahalanobis distance from the abnormal peak to a safe level within 1.05. The system effectively resists external disturbances to the sensing link through a dimensionally consistent adaptive optical power compensation mechanism. Comparative experimental data shows that compared to traditional global proportional current control, this invention reduces the distortion rate of the undamaged wavelength by 89% when subjected to single-point disturbances; and no temperature-induced drive malfunctions were observed within the test temperature range of -10℃ to 60℃.

[0034] Example 5 In a specific implementation of a cross-sensitive decoupling model for ambient temperature thermal disturbance and physical stress micro-bending, addressing the technical problem that the background transmittance of the plastic optical fiber core layer may drift due to thermo-optical effects and thermal expansion, potentially leading to misjudgment and output of additional compensation current by the closed-loop drive circuit, this device constructs a two-parameter cross-sensitive separation model based on multi-wavelength differential absorption characteristics. Specifically, since the influence of ambient temperature on waveguide internal interference and material absorption exhibits broad-spectrum, gradual variation characteristics, its effect on blue light (… ) and red light ( The macroscopic attenuation induced slope is consistent with that of the red light channel; on the contrary, the micro-bending rate perturbation leads to higher-order guided mode radiation leakage. According to Marcuse's fiber bending loss theory, the higher-order mode field diameter of the long-wavelength red light channel is larger and the light confinement is weaker, so it is more sensitive to curvature perturbation than the short-wavelength blue light channel.

[0035] Furthermore, in establishing the location of abnormal nodes... Then, the blue and red light at that location were extracted at specific times. The instantaneous additional logarithmic macroscopic decay change relative to the normal baseline and (All dimensions are...) Establish a scalar model relating environmental temperature changes. (Dimensional °C) and Scalar of Curvature Change (dimension The sensitivity coupling matrix equation is: In the formula, and The thermo-optical attenuation sensitivity of blue and red light to temperature response ( ); and The sensitivity coefficients characterizing the attenuation of blue and red light to deformations with micro-curvature are physically expressed as the attenuation per unit length caused by a unit change in curvature. Their dimensional derivation is: attenuation per unit length (… ) / Curvature change ( = dB. Preferably, the above four sensitivity constants are calibrated by the system in the calibration chamber using a single-variable calibration procedure: first, under constant zero stress, according to... Stepped temperature control chamber from Rise to Obtained by polynomial fitting curve slope ; then maintain Temperature control, using a precision stepper motor to adjust the node curvature from The steps increase to Obtaining the slope of linear regression All of these are stored in the DSP registers. Because and determinant It has a reversible physical basis.

[0036] The controller performs inverse matrix multiplication in real time to calculate the stripping temperature drift. With micro-bending rate variable : Preferably, after decoupling, the system clearly defines the dual-channel collaborative control boundary and data interface: firstly, the calculated curvature variable... The interface mapping equation is used to convert the actual energy deficit ratio of each emission band. ,in The normalized regression coefficients for wavelength deformation attenuation as specified at the factory are the ones used for the initial calibration. To define the sensing range span, the deficit ratio is precisely injected into the discrete PI control law and Kalman observation interface of Example 3; simultaneously, an arbitration superposition rule is defined: if The threshold allows the system to directly look up a table to fine-tune the static reference bias of all wavelength drive currents at the decoupled temperature. Differential PI control performs perturbation dynamic superposition regulation only above this reference bias. This avoids misadjustment of the closed-loop current caused by a sudden increase in external temperature.

[0037] Example 6 In a specific implementation of super-resolution positioning for reconstructing the continuous spatial attenuation distribution of a finite number of discrete sensing points, addressing the technical bottleneck that eight discrete spatial sensing nodes deployed in a link can only provide coarse positioning with adjacent spacing (average span of 2.5m), this device constructs a one-dimensional axial physical field attenuation reconstruction engine based on total variational regularization. Specifically, considering that the attenuation coefficient at the actual physical micro-bending fracture surface exhibits a step-type discontinuity distribution, this invention relaxes the strong smoothness assumption of the function, assuming that the attenuation coefficient at the continuous spatial attenuation distribution is within the entire length of the link. Along the axis coordinate Above, there exists a piecewise smooth one-dimensional spatial decay distribution density function that allows a finite number of step discontinuities. ).

[0038] Furthermore, the system is configured with the measured loss vector calculated from multiple sets of aligned samples as the measurement boundary condition, and a fully variational regularized reconstruction functional formula that takes into account both the step characteristics of discontinuities and noise smoothing is constructed: In the formula, For the objective functional extremum scalar; The dynamic measurement weight of the j-th node is calculated by normalizing the real-time signal-to-noise ratio of each node. The physical distance to the j-th node ( ); The measured cumulative macroscopic loss value (dB) is represented by the second term on the right. The second term on the right is the first-order sparse penalty term of the total variation (TV) regularization, which is used to keenly capture the step-change boundary caused by local cracks or stress fracture. The regularized weights are adaptively obtained by the generalized cross-validation method.

[0039] Specifically, in order to ensure that the functional extremum solution has a numerically definite and reproducible boundary, the processing unit explicitly defines the finite element discretization details: establishing along the one-dimensional full length A one-dimensional discrete finite element mesh of equal length (element length) The nodal physical expansion is performed using first-order piecewise linear Lagrange interpolation basis functions; the boundary conditions are limited to initial steady-state constraints at the output port. The equations for the natural boundary at the end are used. Through finite element discretization and partial differential iteration, a piecewise smooth decay density curve is output across the entire domain. Preferably, in order to improve the positioning accuracy of the step position under discrete numerical solution and avoid the calculation error introduced by numerical differentiation, the system scans a finite differential step quantity difference. The midpoint of the physical unit corresponding to the largest spatial step fault is directly identified as a continuous absolute physical coordinate. : Preferably, for the system integration of super-resolution positioning function and Kalman early warning model in Embodiment 3, the scheme clearly defines the data interface integration logic: when the step size of the reconstructed curve exceeds the abnormal fracture judgment threshold (e.g. The system prioritizes reconfiguring the curve at the worst-case breakpoint. Local small attenuation coefficient maximum value The single-point observation input vector of the Kalman filter in Example 3 is directly injected as a high-optimal measurement value to replace the single-point observation input vector. In this process, the accuracy of the early warning state space convergence is improved. Test data shows that, at a signal-to-noise ratio of... Under operating conditions, this super-resolution positioning process can achieve positioning accuracy within decimeter level (absolute physical deviation) without the need for additional sensing hardware. This effectively solves the problem of super-resolution precise positioning of one-dimensional fiber optic physical fields using low-cost hardware.

[0040] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0041] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-structure optical feedback detection device for an optical fiber illumination system, characterized in that, Include: A multi-channel assembled light source module, comprising multiple independently driven light-emitting chips with different center wavelengths; The integrating sphere combining cavity receives the light beam emitted by the light-emitting chip at its inlet and completes multi-wavelength spatial continuous combining in a polarization-independent manner. A multimode self-collimating optical circulator has a first port, a second port, and a third port. The first port is connected to the output port of the integrating sphere multiplexer, and the second port is connected to the main monitoring link multimode fiber that carries multiple discrete spatial sensing nodes. The optical circulator is configured to transmit the forward multiplexed beam from the first port to the second port with low loss, and to directionally route the random polarization feedback beam returned by the self-collimating link fiber to the third port to achieve directional isolation of the first port. A precision anti-diffraction spatial beam splitting module, connected to the third port, is used to disperse the emitted reverse feedback light and output the optical signals of each wavelength channel through a micro-slit array. The data processing and control motherboard integrates a global real-time hardware clock triggering unit, which is configured to synchronously control the sampling actions of discrete spatial sensing nodes and the spectral sampling of the beam splitting module based on a reference triggering clock, achieve timing alignment, generate a wavelength and spatial joint feature matrix, and perform fault location, optical power closed-loop feedback compensation and early warning based on the matrix.

2. The multi-structure optical feedback detection device for an optical fiber lighting system according to claim 1, characterized in that: A mode field collimation polarization matching channel is provided between the output port of the integrating sphere combining cavity and the first port, and between the second port and the main monitoring link fiber. The channel includes an aspherical aberration-correcting microlens array and a phase-shifting superposition region composed of a spatial birefringent crystal, a non-reciprocal magneto-optical rotator, and a half-wave plate. It is configured to compress, collimate, and decompose the forward diverging multimode beam into orthogonal polarization components, which are then superimposed by polarization and output to the second port in geometric space. It is also configured to cause the randomly polarized feedback light to undergo polarization state exchange and reverse geometric divergence when passing through in the reverse direction. The lateral offset spatial distance of the reverse geometric divergence is greater than the waveguide receiving aperture of the first port to suppress the backflow of feedback light into the light source side.

3. The multi-structure optical feedback detection device for an optical fiber lighting system according to claim 1, characterized in that: The precision anti-diffraction spatial beam splitting module sequentially comprises a collimating lens group, a planar scribed diffraction grating, an imaging focusing lens group, and a metal microstructure beam splitting substrate located at the focal plane. Independent microslits corresponding to the first-order diffraction deflection positions of each emission channel wavelength are formed on the metal microstructure beam-splitting substrate. The physical geometric center distance between two adjacent microslits on the dispersion axis is greater than the sum of the following two physical distances: the first is the sum of the actual geometric opening half-widths of the two adjacent independent microslits, and the second is the diffusion superposition threshold of the principal maximum Airy disk radius determined by the inner diameter of the focusing lens and the two adjacent wavelengths.

4. The multi-structure optical feedback detection device for an optical fiber lighting system according to claim 1, characterized in that: The data processing and control motherboard broadcasts and triggers the distributed sensing nodes to instantaneously latch the light intensity sampling value through the underlying differential bus; at the same time, it is equipped with a high-speed analog-to-digital converter and a dual-port ring first-in-first-out queue, and performs a rectangular window sliding average downsampling operation at the moment the reference trigger is reached to eliminate static dark current noise and generate a downsampling normalized spectral column vector aligned with the current time window. The system converts the light intensity scalar transmitted back from the spatial sensing node into a dimensionless real-time transmission transmittance column vector by deducting the factory-calibrated fixed background loss. Then, the normalized spectral column vector and the transmittance column vector are operated on according to the outer product rule to construct the wavelength and spatial joint characteristic matrix.

5. The multi-structure optical feedback detection device for an optical fiber illumination system according to claim 4, characterized in that: The data processing and control motherboard is configured to expand the wavelength and spatial joint feature matrix into a one-dimensional real-time feature column vector in column priority order, and perform feature matching operations based on Gaussian statistical pattern recognition. Specifically, the deviation of the real-time feature column vector from the expected column vector of various typical working conditions in the standard feature database is calculated, and the weighted quadratic form mapping and square root are performed using the inverse matrix of the positive definite sample covariance matrix of the corresponding pattern to obtain the statistical Mahalanobis distance scalar.

6. The multi-structure optical feedback detection device for an optical fiber lighting system according to claim 1, characterized in that: When the microbending loss state is identified, the data processing and control motherboard locates the abnormal broken node by calculating the maximum physical gradient difference of the transmittance of the spatial sensing node; extracts the entire column of light intensity loss slice vector at the node from the wavelength and spatial joint feature matrix, and calculates the dimensionless relative energy deficit ratio of each emission wavelength relative to the expected value of the lossless value. The system independently adjusts the driving current of each light-emitting chip by executing a discretized differential anti-saturation proportional-integral control law based on the energy deficit ratio. The proportional and integral control gain of the control law is controlled by the decoupling control stability boundary condition. The upper limit of this boundary is jointly determined by the light-emitting channel reference luminous energy, the differential slope of the electro-optic conversion efficiency, and the transmission attenuation coupling coefficient at the break point.

7. The multi-structure optical feedback detection device for an optical fiber lighting system according to claim 1, characterized in that: The system selects a long wavelength channel to construct a continuously evolving dynamic state column vector, which includes the instantaneous equivalent logarithmic decay coefficient, first-order deterioration rate and second-order aging strain acceleration at the positioning node. The system is configured with a time-varying adaptive Kalman recursive model, based on the prior estimation of uniform acceleration state transition, and uses online measurement information and smooth forgetting factor to dynamically update the measurement noise variance, extract the converged second-order acceleration estimate, and use the continuous parabolic trajectory extrapolation formula to calculate the estimated logarithmic decay scalar after a set time delay in the future.

8. The multi-structure optical feedback detection device for an optical fiber lighting system according to claim 6, characterized in that: The system establishes a two-parameter cross-sensitivity separation model; The instantaneous logarithmic macroscopic attenuation change of the short-wavelength and long-wavelength channels at the positioning node is extracted, and a sensitivity coupling matrix is ​​constructed using the pre-calibrated thermo-optical attenuation sensitivity coefficient and deformation attenuation sensitivity coefficient. By performing inverse matrix operations, the mixed attenuation variables are decoupled and separated into pure physical micro-bending rate variables and pure ambient temperature drift. The system only converts the decoupled pure physical micro-bending rate variable into an energy deficit ratio input to the proportional-integral control law for dynamic compensation.

9. The multi-structure optical feedback detection device for an optical fiber lighting system according to claim 8, characterized in that: The system has preset arbitration rules for temperature compensation and dynamic control: When the decoupled ambient temperature drift exceeds the set background thermal compensation threshold, the system directly adjusts the static reference bias current of each wavelength constant current drive circuit according to the temperature lookup table. The proportional-integral control law performs dynamic superposition control of perturbations on the adjusted static reference bias current only for purely physical micro-bends.

10. A multi-structure optical feedback detection device for an optical fiber illumination system according to claim 1, characterized in that: The system is equipped with a super-resolution positioning engine that reconstructs continuous spatial attenuation distribution; The system constructs an objective functional, whose independent variable is a piecewise smooth one-dimensional spatial decay distribution density function. The objective functional includes a data fidelity error term constructed from measured cumulative macroscopic loss values, and a total variational regularization sparsity penalty term composed of first-order spatial derivatives. The system solves the continuous decay density curve that minimizes the functional by finite element discretization partial differential iteration, and scans the maximum value of the spatial finite difference step of the curve to determine the corresponding spatial physical unit coordinates as the continuous physical location coordinates of the fault point.

Images