An argo data driven underwater optical image restoration system (UOIRS) adaptive wavefront reconstruction method and system

Abstract

The application provides an Argo data driven UOIRS underwater light adaptive wavefront reconstruction method and system, and relates to the technical field of underwater visible light communication. The method comprises the following steps: preprocessing Argo buoy profile data to obtain a continuous environmental profile, constructing a depth adaptive improved Nikishov refractive index fluctuation spectrum and generating a mixed turbulence phase screen according to the continuous environmental profile; establishing a tilted non-coaxial propagation geometry model of a transmitting end, a UOIRS and a receiving end, embedding the mixed phase screen into a step-by-step Fourier propagation path, and reconstructing a UOIRS surface complex incident light field; inversely solving an ideal reflection phase according to a phase matching condition, and decomposing the ideal reflection phase into a geometric reflection phase and a turbulence compensation phase; constructing a physical guiding environment perception attention U-Net network, taking the complex incident light field, environmental parameters and turbulence priori diagram as inputs, outputting a spatial variation compensation weight and a residual phase correction amount, generating a PEAU adaptive reflection phase, and quantitatively loading the PEAU adaptive reflection phase to each array element of the UOIRS. The application can reduce residual phase error, improve received light power, and is suitable for underwater visible light communication links.

Description

Technical Field

[0001] This invention relates to the field of underwater visible light communication technology, specifically to an Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method and system. Background Technology

[0002] Underwater visible light communication offers advantages such as high bandwidth, low latency, and strong resistance to electromagnetic interference, making it suitable for underwater sensor networks, submersible collaboration, ocean observation, and short-range high-speed data transmission. However, seawater absorption and scattering, as well as random refractive index disturbances caused by temperature and salinity fluctuations, can lead to beam spreading, drift, scintillation, and wavefront distortion, significantly reducing received optical power and link reliability.

[0003] Existing underwater optical channel models mostly use fixed environmental parameters or equivalent statistical fading to describe the effects of turbulence, making it difficult to reflect the non-stationary characteristics of the temperature and salinity structure of real sea areas as it changes with depth. On the other hand, traditional phase screen generation methods can usually only represent the spectral components that can be resolved by the sampling grid, and their ability to reconstruct low spatial frequency disturbances, wavefront tilt, and beam drift is insufficient. For UOIRS auxiliary links, existing studies are mostly based on ideal reflection or equivalent channel gain, lacking a unified approach to handling the tilted non-coaxial propagation, oblique incidence field mapping, additional optical path difference, and discrete phase constraints of the actual reflector caused by the spatial separation of the transmitter, reflector, and receiver.

[0004] Therefore, a technical solution is needed that can establish a depth-adaptive turbulent phase screen, an end-to-end wave optical propagation model, and an intelligent reflector adaptive wavefront control strategy based on real marine environmental data, in order to improve the reliability and adaptability of underwater tilted non-line-of-sight visible light communication links. Summary of the Invention

[0005] The purpose of this invention is to provide an Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method and system to solve at least one technical problem existing in the prior art.

[0006] On one hand, this invention provides an Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method, the method comprising: Step S1, acquiring an Argo buoy profile dataset of the target sea area, performing standardized preprocessing on the buoy profile dataset, and outputting a continuous temperature profile, a continuous salinity profile, and the corresponding vertical gradient under a unified depth grid; Step S2, constructing a depth-adaptive improved Nikishov seawater refractive index fluctuation power spectrum based on the vertical gradient corresponding to the continuous temperature and salinity profile; Step S3, dividing the underwater optical propagation path into multiple thin turbulent propagation layers, for each layer, retrieving the corresponding turbulent power spectrum of the depth-adaptive improved Nikishov seawater refractive index fluctuation spectrum at the water depth represented by that layer, generating a hybrid turbulent phase screen corresponding to that layer, and synchronously outputting the corresponding turbulent phase modulation function; Step S4, establishing an inclined non-coaxial underwater optical propagation geometric model composed of the transmitter, UOIRS, and receiver, embedding the constructed hybrid turbulent phase screen into the underwater turbulent propagation path, and using the step-by-step Fourier method to calculate the inclined wavefront from the transmitter to the UOIRS. The incident light field evolves, and the oblique incident field in the local coordinate system of UOIRS is reconstructed according to the oblique incident geometry, outputting the complex incident light field on the UOIRS surface; Step S5: Perform phase unwrapping processing on the incident phase distribution of the complex incident light field, establish phase matching constraints that enable the reflected light of each array element to achieve coherent superposition at the receiving end, and obtain the ideal reflection phase by inverse solution, decompose the ideal reflection phase into geometric reflection phase and turbulence compensation phase; Step S6: Construct a physical guided environmental perception attention U-Net network, using the UOIRS complex incident light field, geometric reflection phase, turbulence compensation phase and marine environmental parameters as input, generate spatial variation compensation weights and residual phase corrections, and construct the PEAU adaptive reflection phase based on the spatial variation compensation weights and residual phase corrections; Step S7: According to the preset phase quantization bit number of the UOIRS array element, the continuously distributed PEAU adaptive reflection phase is discretized into a discrete phase with a finite level, and combined with the finite reflection aperture function and reflection efficiency coefficient of UOIRS, the UOIRS reflection function is constructed and loaded into each array element.

[0007] Further, step S1, obtaining the Argo buoy profile dataset of the target sea area, performing standardized preprocessing on the buoy profile dataset, and outputting continuous temperature profiles, continuous salinity profiles, and corresponding vertical gradients under a unified depth grid, includes: Step S101, treating the Argo buoy profile dataset of the target sea area as a discrete dataset composed of depth, temperature, salinity, and pressure: ; in, To ensure an effective number of observation samples, , , and These represent depth, potential temperature or temperature, practical salinity, and pressure, respectively. Step S102: Establish a uniform depth grid within the effective depth range: ; ; in, To standardize depth intervals, To standardize and unify the depth of the j-th layer of the grid, The minimum observation depth within the effective depth range. The maximum observation depth within the effective depth range. To standardize the total number of grid layers, This is a floor function; Step S103: Using the piecewise cubic Hermite interpolation method, the original discrete temperature, salinity, and pressure data are mapped to a uniform depth grid. ; in, This represents the PCHIP interpolation operator. To unify the temperature profile of the interpolated grid, To unify the salinity profile of the interpolated grid, To unify the pressure profile of the interpolated grid; Step S104: Perform Savitzky-Golay smoothing on the interpolated temperature and salinity profiles: ; ; in, and These represent the order of the local fitting polynomial and the length of the smoothing window, respectively. To smooth the continuous temperature profile after noise reduction, To smooth the continuous salinity profile after noise reduction. For Savitzky-Golay smoothing operators; Step S105: Calculate the vertical gradients of the temperature and salinity profiles using central difference: ; ; in, This represents the vertical gradient corresponding to the temperature profile. This represents the vertical gradient corresponding to the salinity profile.

[0008] Furthermore, step S2, constructing a depth-adaptive improved Nikishov seawater refractive index fluctuation power spectrum based on the vertical gradient corresponding to the continuous temperature-salinity profile, includes: Step S201: Based on the continuous temperature profile and salinity profile, and combined with the sensitivity coefficients of seawater refractive index to temperature and salinity, calculate the contributions of the refractive index gradient caused by temperature and salinity respectively: ; ; in, and These represent the sensitivity coefficients of the refractive index to temperature and salinity, respectively. This is the contribution term of the refractive index gradient caused by temperature perturbation. This is the contribution term to the refractive index gradient caused by salinity perturbation; Step S202: Construct regularized temperature-salinity proxy weights based on the vertical gradient corresponding to the continuous temperature-salinity profile. ; in, To regularize the temperature-salt proxy weights, For regularization parameters; Step S203: Construct a depth-varying refractive index structure constant based on the vertical gradient and turbulent external scale corresponding to the continuous temperature-salinity profile: ; in, For calibration coefficients, For the external scale of turbulence that varies with depth, The refractive index is the structural constant. Step S204: Construct a depth-adaptive improved Nikishov seawater refractive index fluctuation power spectrum based on the regularized temperature-salinity proxy weights and the refractive index structure constant varying with depth. ; in, For depth z, space wavenumber The corresponding refractive index fluctuation power spectrum Indicates the transverse spatial wavenumber. This represents the space wavenumber corresponding to the outer scale. , and To control the scaling parameters of high spectral attenuation, It is a natural exponential function.

[0009] Furthermore, the depth-varying turbulent outer scale The expression is: ; in, and These represent the external scales of turbulence in the upper and deep water layers, respectively. A characteristic scale representing how the external scale varies with depth; Lateral spatial wavenumber The expression is: ; in, , Wavenumber components in two orthogonal horizontal directions Space wavenumber corresponding to the outer scale The expression is: .

[0010] Furthermore, in step S3, the underwater optical propagation path is divided into multiple thin turbulent propagation layers. For each layer, the corresponding turbulent power spectrum of the depth-adaptive improved Nikishov seawater refractive index fluctuation spectrum at the water depth represented by that layer is retrieved to generate the corresponding mixed turbulent phase screen for that layer, and the corresponding turbulent phase modulation function for that layer is output synchronously, including: Step S301: Perform plane wave decomposition spectrum sampling on the two-dimensional transverse wavenumber grid of each propagation layer to generate random phase perturbation components that can be distinguished by the sampling grid. Step S302: Introduce multi-level subharmonic sampling in the spatial frequency range below the sampling grid resolution to generate subharmonic low-frequency phase components; Step S303: Using the tilt mode of the Zernike polynomial and the difference between its theoretical covariance matrix and the existing tilt covariance matrix of the phase screen, generate low-order aberration tilt phase components. Step S304: Linearly superimpose the random phase perturbation component, the subharmonic low-frequency phase component, and the low-order aberration tilt phase component, and subtract the spatial average value within the effective area of ​​the superimposed phase screen to remove the piston term, thereby obtaining the mixed turbulence phase screen of this layer. Step S305: Convert the hybrid turbulent phase screen into a complex exponential form of the turbulent phase modulation function.

[0011] Furthermore, in step S4, an inclined non-coaxial underwater light propagation geometric model is established, comprising the transmitter, UOIRS, and receiver. The constructed hybrid turbulence phase screen is embedded into the underwater turbulence propagation path. The evolution of the oblique incident light field from the transmitter to the UOIRS is calculated using the split-step Fourier method. Based on the oblique incident geometry, the oblique incident field in the local coordinate system of the UOIRS is reconstructed, and the complex incident light field on the UOIRS surface, along with its amplitude distribution, intensity distribution, and phase distribution, is output. Step S401: Determine the incident direction and propagation distance from the transmitter to the UOIRS based on the three-dimensional spatial positions of the transmitter, UOIRS, and receiver. Step S402: Construct the initial light field of the emitting end on a transverse plane perpendicular to the incident direction; Step S403: Divide the propagation path from the transmitter to the UOIRS into multiple sub-sections; Step S404: In each step interval, turbulent phase modulation, diffraction propagation and absorption scattering attenuation operations are sequentially applied to the current optical field, wherein the turbulent phase modulation is implemented based on the turbulent phase modulation function of the corresponding layer in step S3; Step S405: Based on the oblique incidence geometry, map the light field reaching UOIRS onto the local coordinate plane of UOIRS, and add the optical path phase caused by oblique incidence to reconstruct the complex incident light field in the local coordinate system of UOIRS.

[0012] Further, in step S5, phase unwrapping processing is performed on the incident phase distribution of the complex incident light field to establish a phase matching constraint that enables coherent superposition of the reflected light from each array element at the receiving end, thus obtaining an ideal reflection phase; the ideal reflection phase is decomposed into a geometric reflection phase and a turbulence compensation phase, including: Step S501: Extract the phase distribution of the complex incident light field and perform two-dimensional phase unwrapping processing to obtain the unwrapped phase; Step S502: Using the center of UOIRS as a reference, obtain the referenced incident unwrapped phase; Step S503: Based on the propagation distance of each UOIRS array element to the receiver, establish the phase matching condition that enables the reflected light from each array element to achieve coherent superposition at the receiver, and solve for the ideal reflection phase. Step S504: Decompose the incident unwrapping phase into the ideal geometric incident phase and the turbulent additional phase, and substitute the ideal reflection phase to obtain the geometric reflection phase and the turbulent compensation phase.

[0013] Further, in step S6, constructing a physically guided environment-aware attention U-Net network, using the UOIRS complex incident light field, geometric reflection phase, turbulence compensation phase, and marine environmental parameters as inputs, generates spatial variation compensation weights and residual phase corrections. Constructing the PEAU adaptive reflection phase based on these spatial variation compensation weights and residual phase corrections includes: Step S601: The amplitude distribution, phase sine, phase cosine, turbulence-added phase, and geometric reflection phase of the complex incident light field constitute a two-dimensional spatial feature input; Step S602: Use marine environmental parameters to construct a one-dimensional environmental feature input, and map it into environmental features through a feature embedding layer; Step S603: Construct a two-dimensional prior map reflecting the spatial distribution of turbulent distortion based on the phase information of the complex incident light field; Step S604: Input the two-dimensional spatial features, the environmental features, and the two-dimensional prior map into the physical-guided environment-aware attention U-Net network. The network extracts multi-scale spatial features through an encoder and performs adaptive weighted fusion of the encoded features based on the environmental features and the two-dimensional prior map through an attention gating mechanism. Step S605: The network decoder compensates for the weights and residual phase corrections based on the spatial changes in the weighted fused feature output. Step S606: Generate the PEAU adaptive reflection phase based on the spatial variation compensation weight, residual phase correction amount, geometric reflection phase and turbulence compensation phase.

[0014] Furthermore, step S7, which involves discretizing and quantizing the continuously distributed PEAU adaptive transmission phase into discrete phases of finite levels based on the preset number of phase quantization bits for the UOIRS array elements, and constructing the UOIRS reflection function and loading it onto each array element by combining the finite reflection aperture function and reflection efficiency coefficient of the UOIRS, includes: Step S701: Perform discrete quantization on the PEAU adaptive reflection phase to obtain the quantized reflection phase; Step S702: Combining the finite aperture function and reflection efficiency coefficient of UOIRS, the quantized reflection phase is constructed into a UOIRS reflection function; Step S703: Multiply the complex incident light field with the UOIRS reflection function point by point to generate the reflected light field after UOIRS modulation.

[0015] Secondly, embodiments of the present invention provide an Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction system. The system employs the aforementioned Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction and includes: an Argo data preprocessing module, suitable for acquiring Argo buoy profile datasets of a target sea area, performing standardized preprocessing on the buoy profile datasets, and outputting continuous temperature profiles, continuous salinity profiles, and corresponding vertical gradients under a unified depth grid; and a depth-adaptive turbulence spectrum construction module, suitable for constructing depth-adaptive improvements based on the vertical gradients corresponding to the continuous temperature and salinity profiles. The module generates a hybrid turbulent phase screen for each layer of the Nikishov seawater refractive index fluctuation power spectrum. It is suitable for dividing the underwater light propagation path into multiple thin turbulent propagation layers. For each layer, it retrieves the corresponding turbulent power spectrum of the depth-adaptive improved Nikishov seawater refractive index fluctuation spectrum at the water depth represented by that layer, generating the corresponding hybrid turbulent phase screen and simultaneously outputting the corresponding turbulent phase modulation function. A module generates a complex incident light field on the surface of a UOIRS (Underwater Oxygen Infrared Reflector), suitable for establishing a tilted non-coaxial underwater light propagation geometric model composed of the transmitter, UOIRS, and receiver. It embeds the constructed hybrid turbulent phase screen into the underwater turbulent propagation path, utilizing… The evolution of the oblique incident light field from the transmitter to the UOIRS is calculated using the split-step Fourier method, and the oblique incident field in the local coordinate system of the UOIRS is reconstructed based on the oblique incident geometry, outputting the complex incident light field on the UOIRS surface. The UOIRS phase matching and decomposition module is suitable for performing phase unwrapping processing on the incident phase distribution of the complex incident light field, establishing phase matching constraints that enable coherent superposition of reflected light from each array element at the receiver, obtaining the ideal reflection phase through inverse solution, and decomposing the ideal reflection phase into a geometric reflection phase and a turbulence compensation phase. The PEAU adaptive reflection phase construction module is suitable for constructing physically guided environmental awareness and attention. The U-Net network takes the UOIRS complex incident light field, geometric reflection phase, turbulence compensation phase, and marine environmental parameters as inputs to generate spatial variation compensation weights and residual phase corrections. Based on these spatial variation compensation weights and residual phase corrections, a PEAU adaptive reflection phase is constructed. The UOIRS reflection function construction module is suitable for discretizing and quantizing the continuously distributed PEAU adaptive reflection phase into a discrete phase with finite levels according to the preset number of phase quantization bits of the UOIRS array elements. Combining the finite reflection aperture function and reflection efficiency coefficient of UOIRS, the UOIRS reflection function is constructed and loaded into each array element.

[0016] Thirdly, embodiments of the present invention also provide an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the above-described Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method.

[0017] Fourthly, embodiments of the present invention also provide a readable storage medium, which, when the instructions in the storage medium are executed by the processor of an electronic device, enables the electronic device to execute the aforementioned Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method.

[0018] The beneficial effects of this invention are as follows: 1. By constructing depth-dependent turbulence parameters using Argo measured profiles, the model can adapt to the temperature, salinity, and thermal structures of different ocean regions, such as the Indian Ocean, the Atlantic Ocean, and the Pacific Ocean. The Argo buoy network provides long-term, continuous global ocean temperature, salinity, and thermal profile observation data. This invention transforms discrete observations into continuous environmental profiles and introduces regularized temperature, salinity, and thermal proxy weights to construct an improved Nikishov spectrum, enabling the phase screen generation to reflect the vertical non-uniformity of the real ocean environment.

[0019] 2. A hybrid phase screen method combining plane wave decomposition, subharmonic compensation, and Zernike low-order reconstruction is employed to improve the consistency of the phase structure function and the ability to represent low-frequency perturbations. The phase screen generated by the power spectrum inversion method suffers from low-frequency energy loss, leading to inaccurate simulation of large-scale effects such as beam drift. This invention recovers large-scale energy through multi-level subharmonic compensation and utilizes Zernike mode compensation to compensate for wavefront tilt, ensuring that the phase structure function matches the theoretical value.

[0020] 3. An end-to-end wave optical model is established, capable of characterizing oblique incidence, additional optical path difference, and actual reflector constraints in UOIRS-assisted tilted non-line-of-sight links. Existing research is mostly based on ideal reflection or equivalent channel gain. This invention fully considers the non-coaxial propagation geometry caused by the spatial separation of the transmitter, UOIRS, and receiver, and establishes a complete propagation model that includes oblique incidence coordinate mapping, additional optical path phase, finite aperture, and discrete quantization constraints.

[0021] 4. The PEAU network can output spatial non-uniformity compensation weights and residual phase correction values, enabling wavefront reconstruction to transform from globally fixed compensation to spatially adaptive compensation. Deep learning-based wavefront correction methods have attracted attention in recent years. This invention integrates physical information into the network through environment embedding, turbulence prior maps, and attention gating mechanisms, achieving refined control of spatial non-uniform turbulence distortion and reducing residual phase errors. Attached Figure Description

[0022] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0023] Figure 1 This is a schematic diagram of the Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method provided in Embodiment 1 of the present invention.

[0024] Figure 2This is a schematic diagram of the process of converting Argo data from discrete observations into a continuous profile, as provided in Embodiment 1 of the present invention.

[0025] Figure 3 This is a local turbulent phase screen and cumulative propagation intensity distribution map on different propagation layers provided in Embodiment 1 of the present invention.

[0026] Figure 4 This is a comparison diagram of the empirical phase structure function and the theoretical phase structure function obtained by different phase screen generation methods provided in Embodiment 1 of the present invention.

[0027] Figure 5 This is a schematic diagram of a UOIRS-assisted tilted underwater optical link structure provided in Embodiment 1 of the present invention.

[0028] Figure 6 This is a diagram of a physical-guided environment-aware attention U-Net adaptive wavefront reconstruction network structure provided in Embodiment 1 of the present invention.

[0029] Figure 7a It is a spatial compensation weight map obtained by fixing uniform compensation weights.

[0030] Figure 7b It is a spatially compensated weight map obtained from a regular U-Net.

[0031] Figure 7c This is the spatial compensation weight map obtained by PEAU in this invention.

[0032] Figure 8a It is an absolute phase error heatmap obtained by fixing uniform compensation weights.

[0033] Figure 8b This is a heatmap of the absolute phase error obtained from a standard U-Net.

[0034] Figure 8c This is the absolute phase error heatmap obtained by PEAU in this invention.

[0035] Figure 9a It is the initial light intensity distribution at the transmitter (TX).

[0036] Figure 9b The light intensity distribution at the receiver of the direct link shows obvious attenuation and dispersion.

[0037] Figure 9c The light intensity distribution at the fixed UOIRS receiver was slightly improved, but it failed to focus effectively.

[0038] Figure 9d It features adaptive UOIRS receiver light intensity distribution, achieving efficient beam focusing and significant light intensity recovery.

[0039] Figure 10 This is a schematic diagram of an Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction system provided in Embodiment 2 of the present invention.

[0040] Figure 11 This is a partial block diagram of the electronic device provided in Embodiment 3 of the present invention. Detailed Implementation

[0041] Before discussing the exemplary embodiments in more detail, it should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations as sequential processes, many of these operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but may also have additional steps not included in the figures. The process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0042] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0043] The present invention will now be described in detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0044] Example 1 The specific implementation method is as follows: like Figure 1 The diagram shown is a flowchart of an Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method provided by the present invention.

[0045] As an example, the method includes: Step S1, acquiring the Argo buoy profile dataset of the target sea area, performing standardized preprocessing on the buoy profile dataset, and outputting continuous temperature profiles, continuous salinity profiles, and corresponding vertical gradients under a unified depth grid; Step S2, constructing a depth-adaptive improved Nikishov seawater refractive index fluctuation power spectrum based on the vertical gradient corresponding to the continuous temperature and salinity profiles; Step S3, dividing the underwater light propagation path into multiple thin turbulent propagation layers, and for each layer, retrieving the corresponding turbulent power spectrum of the depth-adaptive improved Nikishov seawater refractive index fluctuation spectrum at the water depth represented by that layer, generating the corresponding hybrid turbulent phase screen for that layer, and simultaneously outputting the corresponding turbulent phase modulation function for that layer; Step S4, establishing an inclined non-coaxial underwater light propagation geometric model composed of the transmitter, UOIRS, and receiver, embedding the constructed hybrid turbulent phase screen into the underwater turbulent propagation path, calculating the evolution of the oblique incident light field from the transmitter to the UOIRS using the split-step Fourier method, and performing calculations based on the oblique incident geometric relationship. Reconstructing the oblique incident field in the local coordinate system of UOIRS, outputting the complex incident light field on the UOIRS surface; Step S5: Performing phase unwrapping processing on the incident phase distribution of the complex incident light field, establishing a phase matching constraint that enables the reflected light of each array element to achieve coherent superposition at the receiving end, inversely solving to obtain the ideal reflection phase, and decomposing the ideal reflection phase into a geometric reflection phase and a turbulence compensation phase; Step S6: Constructing a physically guided environmental awareness attention U-Net network, using the UOIRS complex incident light field, geometric reflection phase, turbulence compensation phase, and marine environmental parameters as inputs, generating spatial variation compensation weights and residual phase corrections, and constructing a PEAU adaptive reflection phase based on the spatial variation compensation weights and residual phase corrections; Step S7: According to the preset phase quantization bit number of the UOIRS array element, discretizing and quantizing the continuously distributed PEAU adaptive reflection phase into a discrete phase with finite levels, combining the finite reflection aperture function and reflection efficiency coefficient of UOIRS, constructing the UOIRS reflection function and loading it onto each array element.

[0046] In some feasible implementations, step S1, obtaining the Argo buoy profile dataset of the target sea area, performing standardized preprocessing on the buoy profile dataset, and outputting continuous temperature profiles, continuous salinity profiles, and corresponding vertical gradients under a unified depth grid, includes: Step S101: Record the Argo buoy profile dataset of the target sea area as a discrete dataset consisting of depth, temperature, salinity, and pressure: ; in, To ensure an effective number of observation samples, , , and These represent depth, potential temperature or temperature, practical salinity, and pressure, respectively. Step S102: Establish a uniform depth grid within the effective depth range: ; ; in, To standardize depth intervals, To standardize and unify the depth of the j-th layer of the grid, The minimum observation depth within the effective depth range. The maximum observation depth within the effective depth range. To standardize the total number of grid layers, This is a floor function; Step S103: Using the piecewise cubic Hermite interpolation method, the original discrete temperature, salinity, and pressure data are mapped to a uniform depth grid. ; in, This represents the PCHIP interpolation operator. To unify the temperature profile of the interpolated grid, To unify the salinity profile of the interpolated grid, To unify the pressure profile of the interpolated grid; Step S104: Perform Savitzky-Golay smoothing on the interpolated temperature and salinity profiles: ; ; in, and These represent the order of the local fitting polynomial and the length of the smoothing window, respectively. To smooth the continuous temperature profile after noise reduction, To smooth the continuous salinity profile after noise reduction. For Savitzky-Golay smoothing operators; Step S105: Calculate the vertical gradients of the temperature and salinity profiles using central difference: ; ; in, This represents the vertical gradient corresponding to the temperature profile. This represents the vertical gradient corresponding to the salinity profile.

[0047] Thus, the non-uniformly distributed discrete Argo observation data is transformed into a continuous temperature profile. Continuous salinity profile And its vertical gradient. A schematic diagram illustrating the process of converting Argo data from discrete observations into continuous profiles of the Indian Ocean, Atlantic Ocean, and Pacific Ocean. Figure 2 As shown.

[0048] In some feasible implementations, step S2, which involves constructing a depth-adaptive improved Nikishov seawater refractive index fluctuation power spectrum based on the vertical gradient corresponding to the continuous temperature-salinity profile, includes: Step S201: Based on the continuous temperature profile and salinity profile, and combined with the sensitivity coefficients of seawater refractive index to temperature and salinity, calculate the contributions of the refractive index gradient caused by temperature and salinity respectively: ; ; in, and These represent the sensitivity coefficients of the refractive index to temperature and salinity, respectively. This is the contribution term of the refractive index gradient caused by temperature perturbation. This is the contribution term to the refractive index gradient caused by salinity perturbation; Step S202: Construct regularized temperature-salinity proxy weights based on the vertical gradient corresponding to the continuous temperature-salinity profile. ; in, To regularize the temperature-salt proxy weights, For regularization parameters; Step S203: Construct a depth-varying refractive index structure constant based on the vertical gradient and turbulent external scale corresponding to the continuous temperature-salinity profile: ; in, For calibration coefficients, For the external scale of turbulence that varies with depth, The refractive index is the structural constant. Step S204: Construct a depth-adaptive improved Nikishov seawater refractive index fluctuation power spectrum based on the regularized temperature-salinity proxy weights and the refractive index structure constant varying with depth. ; in, For depth z, space wavenumber The corresponding refractive index fluctuation power spectrum Indicates the transverse spatial wavenumber. This represents the space wavenumber corresponding to the outer scale. , and To control the scaling parameters of high spectral attenuation, It is a natural exponential function.

[0049] Preferably, the depth-varying turbulent outer scale The expression is: ; in, and These represent the external scales of turbulence in the upper and deep water layers, respectively. A characteristic scale representing how the external scale varies with depth; Lateral spatial wavenumber The expression is: ; in, , Wavenumber components in two orthogonal horizontal directions Space wavenumber corresponding to the outer scale The expression is: .

[0050] That is, the spectrum contains main items, including One revision item and containing The superposition of the second-order correction terms can reflect the influence of the temperature-salinity dual diffusion effect on the refractive index fluctuation spectrum.

[0051] In some feasible implementations, step S3, dividing the underwater optical propagation path into multiple thin turbulent propagation layers, and for each layer, retrieving the corresponding turbulent power spectrum of the depth-adaptive improved Nikishov seawater refractive index fluctuation spectrum at the water depth represented by that layer, generating the corresponding mixed turbulent phase screen for that layer, and simultaneously outputting the corresponding turbulent phase modulation function for that layer, includes: The underwater light propagation path is divided into The propagation layer, the first The representative depth and propagation thickness of the layer are denoted as follows: and Based on the turbulent power spectrum corresponding to this layer. The plane wave decomposition phase component, the subharmonic low-frequency compensation phase component, and the Zernike low-order aberration phase component are generated respectively.

[0052] Preferably, step S301 includes: randomly sampling the resolvable spatial frequencies using a plane wave decomposition method, the... The random phase perturbation components of the layer are represented as follows: ; in, , Let be a discrete spatial wavenumber vector, and let the phase spectral coefficient be... ; in, A complex Gaussian random variable with zero mean and unit variance. The optical wavenumber in water.

[0053] To compensate for the low-frequency energy that cannot be covered by conventional spectrum sampling, a method is introduced... Level subharmonic sampling, its first The frequency interval of each stage is: ; in, This is the physical side length of the phase screen.

[0054] Preferably, step S302 includes: the subharmonic low-frequency phase component is represented as: ; Among them, coefficient The variance is determined by the turbulence spectrum at the corresponding low-frequency position, and the zero-frequency component is excluded to avoid introducing the piston term.

[0055] To further recover the large-scale wavefront tilt that is not fully characterized by plane wave decomposition and subharmonic methods, a residual covariance is constructed based on the difference between the theoretical tilt covariance and the existing phase screen tilt covariance: ; Decompose the residual covariance: ; And generate random coefficients for two orthogonal tilt modes: ; in, and These are mutually independent standard Gaussian random variables.

[0056] Preferably, step S303 includes: the corresponding Zernike low-order aberration tilt phase component is: ; in, and These represent two orthogonal tilt modes on the normalized circular aperture.

[0057] Preferably, step S304 includes: the first The mixed turbulent phase screen of each propagation layer is represented as: ; Preferably, step S305 includes: removing the piston term from the hybrid phase screen to eliminate the overall phase shift that does not affect the optical field propagation characteristics. ; in, This indicates that spatial averaging is performed within the effective phase screen area.

[0058] The corresponding turbulent phase modulation function is: ; This hybrid phase screen describes the main spatial frequency components through plane wave decomposition, compensates for the lack of low-frequency energy using the subharmonic method, and recovers the large-scale wavefront tilt through Zernike mode, thereby improving the low-frequency statistical accuracy and spatial structural integrity of the underwater turbulent phase screen. Figure 3 The local turbulent phase screen and cumulative propagation intensity distribution of the method of the present invention on different propagation layers are shown. It can be seen that as the propagation distance increases, the light spot shows obvious intensity fluctuations and expansion. Figure 4 By comparing the empirical phase structure function obtained from different phase screen generation methods with the theoretical value, the hybrid method proposed in this invention almost coincides with the theoretical curve across the entire spatial distance range, verifying its high accuracy in statistical characterization.

[0059] In some feasible implementations, step S4, establishing an inclined non-coaxial underwater light propagation geometric model consisting of the transmitter, UOIRS, and receiver, embedding the constructed hybrid turbulence phase screen into the underwater turbulence propagation path, calculating the evolution of the oblique incident light field from the transmitter to the UOIRS using the split-step Fourier method, and reconstructing the oblique incident field in the local coordinate system of the UOIRS based on the oblique incident geometry, outputting the complex incident light field on the UOIRS surface and its amplitude distribution, intensity distribution, and phase distribution, includes: Preferably, step S401 includes: the input for step S4 includes the transmitter position. UOIRS Central Location Location of the receiving end UOIRS surface normal vector Transmission power Launch aperture radius wavelength of light absorption coefficient Scattering coefficient , the number of propagation segments and the hybrid phase screen sequence generated in step S3 ; Based on the three-dimensional positions of the transmitter and the center of UOIRS, calculate the propagation distance and unit incident direction:

[0060] The angle between the incident direction and the surface normal vector of UOIRS is: ; When the incident beam is in the global When in a plane, its unit incident direction can be expressed as: ; by Establish a local propagation coordinate system for the longitudinal propagation direction. Its horizontal basis vectors are: ; ; Any point in the propagation path can be represented as: ; Preferably, step S402 includes: constructing an initial complex optical field of the transmitting end on a local lateral coordinate plane based on the transmitting power and the transmitting aperture. ;in For transmission power, The radius of the emission aperture; The propagation distance from the transmitter to the UOIRS center is divided into... There are 3 propagation segments, each with a length of 1. ; No. The hybrid phase screen loaded in each propagation segment is: ; The three parts are the plane wave decomposition phase, the subharmonic low-frequency compensation phase, and the Zernike tilt compensation phase.

[0061] Let the underwater extinction coefficient be , where a is the absorption coefficient and b is the scattering coefficient.

[0062] Preferably, steps S403 and S404 include: for the first In each propagation segment, the input light field is first phase-modulated using a hybrid phase screen: ; Subsequently, the diffraction propagation was calculated using the angular spectral propagation operator, and absorption scattering attenuation was introduced: ; For the light field after the propagation of the first segment, For the m-th segment, a hybrid phase screen is loaded. For angular spectrum propagation operators, This is the length of a single propagation segment.

[0063] The angular spectrum propagation operator is represented as: ; It is a two-dimensional Fourier transform; The angular spectral transfer function is: ; The underwater light wavenumber is: ; The refractive index of seawater, Wavelength in vacuum; Following the above relationships step by step, after... After propagation segments, the complex optical field of the terminal cross section perpendicular to the incident direction is obtained: ; UOIRS Local Coordinates With propagation coordinates The following conditions must be met: ; in, This represents the lateral projection produced by oblique incidence. This represents the additional optical path difference corresponding to different positions on the UOIRS surface.

[0064] Preferably, step S405 includes: the light field of the terminal cross-section projected onto the UOIRS surface is as follows: ; The additional optical path difference and its corresponding geometric phase are as follows: ; ; The complex incident light field on the UOIRS surface is represented as: ; Its amplitude, intensity, and phase distribution are as follows: ; ; .

[0065] Therefore, step S4 ultimately outputs the complex incident light field defined on the physical surface of UOIRS. and the corresponding amplitude distribution Light intensity distribution and phase distribution The output comprehensively considers hybrid phase screen perturbation, diffraction propagation, absorption and scattering attenuation, oblique incidence coordinate projection, and additional optical path phase, and serves as the input for subsequent UOIRS reflection phase design and adaptive wavefront compensation. A schematic diagram of its UOIRS-assisted tilted underwater optical link structure is shown below. Figure 5 As shown.

[0066] In some feasible implementations, step S5, performing phase unwrapping processing on the incident phase distribution of the complex incident light field to establish a phase-matching constraint that enables coherent superposition of the reflected light from each array element at the receiving end, yields an ideal reflection phase; decomposing the ideal reflection phase into a geometric reflection phase and a turbulence-compensated phase includes: The complex incident light field on the UOIRS surface is obtained from step S4: ; Preferably, step S501 includes: firstly, extracting the incident phase and performing two-dimensional phase unwrapping processing: ; Preferably, step S502 includes: using the UOIRS center as a reference, obtaining the referenced incident unwrapped phase: ; Let UOIRS be the surface element. The distance to the receiving end is: ; Preferably, step S503 includes: in order to ensure that the reflected light from each surface element is coherently superimposed at the receiving end, a phase matching condition is established with the center of UOIRS as the phase reference: ; Therefore, the ideal reflection phase can be obtained by inverse solving: ; Taking the center of UOIRS as the phase reference, it can be written as: ; Preferably, step S504 includes: decomposing the incident unwrapped phase into an ideal geometrical incident phase. and turbulent additional phase : ; The geometric reflection phase and the turbulence compensation phase are respectively: ; .

[0067] In some feasible implementations, step S6, constructing a physically guided environment-aware attention U-Net network, using the UOIRS complex incident light field, geometric reflection phase, turbulence compensation phase, and marine environmental parameters as inputs, generates spatial variation compensation weights and residual phase corrections. Constructing the PEAU adaptive reflection phase based on these spatial variation compensation weights and residual phase corrections includes: Step S6 uses the UOIRS complex incident light field obtained in step S4, the geometric reflection phase and turbulence compensation phase obtained in step S5, and ocean environmental parameters as inputs to construct a physics-guided environmental awareness attention U-Net. The network includes a two-dimensional physics input branch, an environmental parameter input branch, and a turbulence prior guidance branch, and outputs spatial compensation weights. and residual phase correction amount .

[0068] Preferably, step S601 includes: obtaining the complex incident light field of the UOIRS surface from step S4. ; in, ; represents the amplitude of the incident light field. The referenced incident unwrapped phase.

[0069] Because of the phase and There is a periodic jump at this point, and the phase is represented in sine and cosine form: ; ; The ideal turbulent-free injection phase is denoted as Then the additional phase of turbulence is: ; The corresponding turbulence compensation phase is: ; Therefore, the two-dimensional physical field input of PEAU is expressed as .

[0070] Preferably, step S602 includes: constructing a one-dimensional environment feature vector based on the Argo profile data and propagation geometry parameters. ; in, For temperature, Salinity For pressure, For depth, For turbulence intensity, and These are the external and internal scales of turbulence, respectively. The propagation distance from the transmitter to the UOIRS. The angle of incidence is denoted as .

[0071] First, the environmental parameters with different dimensions are normalized: ; in, and These represent the mean vector and scale matrix of the environmental parameters, respectively.

[0072] The normalized environment vector is then input into the multilayer perceptron: ; Its expanded form can be written as: ; in, It is a non-linear activation function.

[0073] Copy the environment embedding vector to each spatial location: ; Finally, the main input to the network is obtained by concatenating two-dimensional physical field features and two-dimensional environmental features: .

[0074] Preferably, step S603 includes: first calculating the spatial gradient magnitude of the turbulent phase: ; Weighted combination of turbulence phase magnitude and spatial gradient: ; in, These are gradient weights; Normalizing it yields the turbulence prior diagram: ; in, The larger the value, the stronger the turbulent phase distortion or the more drastic the spatial variation at that location.

[0075] Preferably, step S604 includes: inputting the network... Input the U-Net encoder to obtain the first... Layer coding features: ; During the decoding process, the encoded features, decoded features, environment embedding, and turbulence prior map are all input into the attention gate: ; in, For the Sigmoid function, For the first Layer attention coefficient.

[0076] The gated encoded features are: ; in, This represents element-wise multiplication; Concatenate it with the upsampled decoded features: ; The final decoded features are obtained through multi-scale encoding, attention gating, and layer-by-layer decoding. .

[0077] Preferably, step S605 includes: using Convolution maps the final decoded features into two output branches: ; Applying the Sigmoid function to the first output branch yields the spatial variation compensation weights for the network predictions: ; To further incorporate turbulence prior constraints, the network learning results can be fused with the turbulence prior map: ; in, These are the prior fusion coefficients, therefore ; when A larger value indicates that the region requires stronger turbulent phase compensation; a smaller value indicates that the region mainly uses geometric reflection phase.

[0078] Applying the hyperbolic tangent function to the second output branch yields the residual phase correction: ; in, For the maximum residual phase correction range, therefore .

[0079] Preferably, step S606 includes: Based on the geometric reflection phase obtained in step S5 and turbulence compensation phase Construct the PEAU reflection phase: ; in, ; ; Therefore, the PEAU reflection phase can be further expanded as follows: ; The first term is used to compensate for the ideal tilted incident phase; the second term is used to achieve coherent focusing of the reflected light from the surface element at the receiving end; the third term is used to adaptively compensate for turbulence distortion at different spatial locations; and the fourth term is used to correct residual errors caused by factors such as analytical phase model, finite aperture, phase screen discretization, and phase quantization.

[0080] Preferably, constructing the training objective and loss function includes: Let the ideal target reflection phase be: ; Based on the turbulence prior diagram and the global compensation weight obtained in step S5 A spatial compensation weight objective can be constructed: ;

[0081] in, Used to control the intensity of spatial non-uniformity compensation.

[0082] The corresponding residual phase target is: ; in, Used to limit the phase to Inside.

[0083] Phase reconstruction loss is defined as: ; Spatial compensation weight loss is defined as: ; The residual phase smoothing loss is defined as: ; The total network loss is: ; in, and These are the weight coefficients for spatial compensation weight loss and smoothing loss, respectively.

[0084] The PEAU adaptive wavefront reconstruction network structure proposed in this invention is as follows: Figure 6 As shown in the figure, the complete architecture of the encoder, decoder, skip connections, context embedding layer, attention gating unit, and dual-branch output layer is clearly illustrated. Figures 7a to 7c By comparing the spatial compensation weight distributions obtained by fixed uniform compensation weights, ordinary U-Net, and the PEAU of this invention, it can be seen that the weights of PEAU exhibit a clear spatial hierarchy—the weights are highest in the lower central region and gradually decrease towards the periphery, which is highly consistent with the turbulence prior map, realizing the spatial adaptive strategy of "focusing on compensation where turbulence is strong". Figures 8a to 8c The absolute phase error heatmap shows that the residual phase error after PEAU compensation (uniform blue throughout the image) is significantly lower than that after fixed compensation and ordinary U-Net, verifying the effectiveness of the physical guidance mechanism.

[0085] In some feasible implementations, step S7 includes: for Bit UOIRS, phase quantization interval is: ; The quantized PEAU reflection phase is: ; Considering both finite aperture and reflection efficiency, the PEAU reflection function is: ; The reflected light field after UOIRS modulation is as follows: .

[0086] Comparison of light intensity distribution at the transmitter and receiver under different transmission schemes is shown in the figure below. Figures 9a to 9d As shown, after PEAU compensation, the beam pattern is closest to that of the transmitter—the main lobe is concentrated, circularly symmetrical, and there are no visible side lobes. The central peak intensity reaches the highest level, indicating that the reflected wavefront has been precisely leveled, achieving the most efficient coherent superposition at the receiver.

[0087] In summary, this invention, starting from real ocean data, uses a physical optical propagation model as its core and deep learning technology as its optimization engine to propose a UOIRS wavefront reconstruction method that combines physical interpretability and data adaptability. Experimental results demonstrate that this invention can effectively reduce the residual phase error of underwater tilted links, increase received optical power, and reduce the probability of interruption and the bit error rate, providing a new technical path for high-speed and reliable underwater visible light communication in complex ocean environments.

[0088] Example 2 Please see Figure 10 This embodiment provides a schematic diagram of the structure of an Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction system.

[0089] As an example, the system is implemented using the Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction described in Example 1. The system includes: Argo data preprocessing module 1 is suitable for acquiring Argo buoy profile datasets of target sea areas, performing standardized preprocessing on the buoy profile datasets, and outputting continuous temperature profiles, continuous salinity profiles, and corresponding vertical gradients under a unified depth grid. The depth-adaptive turbulence spectrum construction module 2 is suitable for constructing a depth-adaptive improved Nikishov seawater refractive index fluctuation power spectrum based on the vertical gradient corresponding to the continuous temperature and salinity profile. The hybrid phase screen generation module 3 is suitable for dividing the underwater light propagation path into multiple thin turbulent propagation layers. For each layer, the corresponding turbulent power spectrum of the depth-adaptive improved Nikishov seawater refractive index fluctuation spectrum at the water depth represented by the layer is retrieved to generate the hybrid turbulent phase screen corresponding to the layer, and the corresponding turbulent phase modulation function is output synchronously. The complex incident light field generation module 4 on the surface of UOIRS is suitable for establishing an inclined non-coaxial underwater light propagation geometric model consisting of the transmitter, UOIRS and receiver. The constructed hybrid turbulence phase screen is embedded in the underwater turbulence propagation path. The evolution of the oblique incident light field from the transmitter to UOIRS is calculated using the split-step Fourier method. The oblique incident field in the local coordinate system of UOIRS is reconstructed according to the oblique incident geometry, and the complex incident light field on the surface of UOIRS is output. UOIRS phase matching and decomposition module 5 is suitable for performing phase unwrapping processing on the incident phase distribution of the complex incident light field, establishing a phase matching constraint that enables the reflected light of each array element to achieve coherent superposition at the receiving end, inversely solving to obtain the ideal reflection phase, and decomposing the ideal reflection phase into a geometric reflection phase and a turbulence compensation phase. The PEAU adaptive reflection phase construction module 6 is suitable for constructing a physically guided environmental awareness attention U-Net network. It takes the UOIRS complex incident light field, geometric reflection phase, turbulence compensation phase and marine environmental parameters as inputs to generate spatial variation compensation weights and residual phase corrections. Based on the spatial variation compensation weights and residual phase corrections, the PEAU adaptive reflection phase is constructed. UOIRS reflection function construction module 7 is suitable for discretizing and quantizing the continuously distributed PEAU adaptive reflection phase into a discrete phase with a finite level according to the preset phase quantization bit number of the UOIRS array element, and constructing the UOIRS reflection function and loading it onto each array element by combining the finite reflection aperture function and reflection efficiency coefficient of UOIRS.

[0090] It is not difficult to see that this embodiment is a system implementation corresponding to the first embodiment, and this embodiment can be implemented in conjunction with the first embodiment. The relevant technical details mentioned in the first embodiment are still valid in this embodiment, and will not be repeated here to reduce repetition. Accordingly, the relevant technical details mentioned in this embodiment can also be applied to the first embodiment.

[0091] It is worth mentioning that all modules involved in this embodiment are logical units. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. Furthermore, to highlight the innovative aspects of this invention, this embodiment does not introduce units that are not closely related to solving the technical problem proposed by this invention; however, this does not mean that other units are absent from this embodiment.

[0092] Example 3 Please see Figure 11 The present invention also provides an electronic device, including: a memory and a processor; the memory stores at least one program instruction; the processor loads and executes the at least one program instruction to implement the Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method provided in Embodiment 1.

[0093] The memory 702 and processor 701 are connected via a bus, which may include any number of interconnecting buses and bridges, connecting various circuits of one or more processors 701 and memory 702 together. The bus may also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. A bus interface provides an interface between the bus and the transceiver. The transceiver may be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by processor 701 is transmitted over a wireless medium via an antenna, which further receives data and transmits it to processor 701.

[0094] Processor 701 is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory 702 can be used to store data used by processor 701 during operation.

[0095] Example 4 This invention also proposes a storage medium storing an Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method. When the Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction program is executed by a processor, it implements the steps of the Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method as described above. Since this storage medium adopts all the technical solutions of all the above embodiments, it possesses at least all the beneficial effects brought about by the technical solutions of the above embodiments, which will not be elaborated further here.

[0096] The above descriptions are merely embodiments of the present invention. Commonly known structures and characteristics are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are aware of all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, based on the guidance provided in this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. An Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method, characterized in that, The method includes: Step S1: Obtain the Argo buoy profile dataset of the target sea area, perform standardization preprocessing on the buoy profile dataset, and output continuous temperature profile, continuous salinity profile and corresponding vertical gradient under a unified depth grid. Step S2: Construct a depth-adaptive improved Nikishov seawater refractive index fluctuation power spectrum based on the vertical gradient corresponding to the continuous temperature-salinity profile; Step S3: Divide the underwater light propagation path into multiple thin turbulent propagation layers. For each layer, retrieve the corresponding turbulent power spectrum of the depth-adaptive improved Nikishov seawater refractive index fluctuation spectrum at the water depth represented by the layer, generate the corresponding mixed turbulent phase screen for that layer, and simultaneously output the corresponding turbulent phase modulation function for that layer. Step S4: Establish an inclined non-coaxial underwater light propagation geometric model consisting of the transmitter, UOIRS and receiver. Embed the constructed hybrid turbulence phase screen into the underwater turbulence propagation path. Calculate the evolution of the oblique incident light field from the transmitter to the UOIRS using the split-step Fourier method. Reconstruct the oblique incident field in the local coordinate system of the UOIRS according to the oblique incident geometric relationship, and output the complex incident light field on the surface of the UOIRS. Step S5: Perform phase unwrapping processing on the incident phase distribution of the complex incident light field, establish a phase matching constraint that enables the reflected light of each array element to achieve coherent superposition at the receiving end, solve the inverse to obtain the ideal reflection phase, and decompose the ideal reflection phase into the geometric reflection phase and the turbulence compensation phase. Step S6: Construct a physically guided environment-aware attention U-Net network, using the UOIRS complex incident light field, geometric reflection phase, turbulence compensation phase, and marine environmental parameters as inputs to generate spatial variation compensation weights and residual phase corrections. Based on the spatial variation compensation weights and residual phase corrections, construct a PEAU adaptive reflection phase. Step S7: Based on the preset number of phase quantization bits of the UOIRS array elements, the continuously distributed PEAU adaptive reflection phase is discretized into discrete phases with finite levels. Combined with the finite reflection aperture function and reflection efficiency coefficient of UOIRS, the UOIRS reflection function is constructed and loaded into each array element.

2. The Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method according to claim 1, characterized in that, Step S1 involves acquiring the Argo buoy profile dataset for the target sea area, performing standardized preprocessing on the buoy profile dataset, and outputting continuous temperature profiles, continuous salinity profiles, and corresponding vertical gradients under a unified depth grid, including: Step S101: Record the Argo buoy profile dataset of the target sea area as a discrete dataset consisting of depth, temperature, salinity, and pressure: ； in, To ensure an effective number of observation samples, , , and These represent depth, potential temperature or temperature, practical salinity, and pressure, respectively. Step S102: Establish a uniform depth grid within the effective depth range: ； ； in, To standardize depth intervals, To standardize and unify the depth of the j-th layer of the grid, The minimum observation depth within the effective depth range. The maximum observation depth within the effective depth range. To standardize the total number of grid layers, This is a floor function; Step S103: Using the piecewise cubic Hermite interpolation method, the original discrete temperature, salinity, and pressure data are mapped to a uniform depth grid. ； in, This represents the PCHIP interpolation operator. To unify the temperature profile of the interpolated grid, To unify the salinity profile of the interpolated grid, To unify the pressure profile of the interpolated grid; Step S104: Perform Savitzky-Golay smoothing on the interpolated temperature and salinity profiles: ； ； in, and These represent the order of the local fitting polynomial and the length of the smoothing window, respectively. To smooth the continuous temperature profile after noise reduction, To smooth the continuous salinity profile after noise reduction. For Savitzky-Golay smoothing operators; Step S105: Calculate the vertical gradients of the temperature and salinity profiles using central difference: ； ； in, This represents the vertical gradient corresponding to the temperature profile. This represents the vertical gradient corresponding to the salinity profile.

3. The Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method according to claim 2, characterized in that, Step S2, constructing a depth-adaptive improved Nikishov seawater refractive index fluctuation power spectrum based on the vertical gradient corresponding to the continuous temperature-salinity profile, includes: Step S201: Based on the continuous temperature profile and salinity profile, and combined with the sensitivity coefficients of seawater refractive index to temperature and salinity, calculate the contributions of the refractive index gradient caused by temperature and salinity respectively: ； ； in, and These represent the sensitivity coefficients of the refractive index to temperature and salinity, respectively. This is the contribution term of the refractive index gradient caused by temperature perturbation. This is the contribution term to the refractive index gradient caused by salinity perturbation; Step S202: Construct regularized temperature-salinity proxy weights based on the vertical gradient corresponding to the continuous temperature-salinity profile. ； in, To regularize the temperature-salt proxy weights, For regularization parameters; Step S203: Construct a depth-varying refractive index structure constant based on the vertical gradient and turbulent external scale corresponding to the continuous temperature-salinity profile: ； in, For calibration coefficients, For the external scale of turbulence that varies with depth, The refractive index is the structural constant. Step S204: Construct a depth-adaptive improved Nikishov seawater refractive index fluctuation power spectrum based on the regularized temperature-salinity proxy weights and the refractive index structure constant varying with depth. ； in, For depth z, space wavenumber The corresponding refractive index fluctuation power spectrum Indicates the transverse spatial wavenumber. This represents the space wavenumber corresponding to the outer scale. , and To control the scaling parameters of high spectral attenuation, It is a natural exponential function.

4. The Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method according to claim 3, characterized in that, The depth-varying turbulent outer scale The expression is: ； in, and These represent the external scales of turbulence in the upper and deep water layers, respectively. A characteristic scale representing how the external scale varies with depth; Lateral spatial wavenumber The expression is: ； in, , Wavenumber components in two orthogonal horizontal directions Space wavenumber corresponding to the outer scale The expression is: 。 5. The Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method according to claim 1, characterized in that, Step S3 involves dividing the underwater optical propagation path into multiple thin turbulent propagation layers. For each layer, the corresponding turbulent power spectrum of the depth-adaptive improved Nikishov seawater refractive index fluctuation spectrum at the water depth represented by that layer is retrieved to generate the mixed turbulent phase screen corresponding to that layer. Simultaneously, the turbulent phase modulation function corresponding to that layer is output, including: Step S301: Perform plane wave decomposition spectrum sampling on the two-dimensional transverse wavenumber grid of each propagation layer to generate random phase perturbation components that can be distinguished by the sampling grid. Step S302: Introduce multi-level subharmonic sampling in the spatial frequency range below the sampling grid resolution to generate subharmonic low-frequency phase components; Step S303: Using the tilt mode of the Zernike polynomial and the difference between its theoretical covariance matrix and the existing tilt covariance matrix of the phase screen, generate low-order aberration tilt phase components. Step S304: Linearly superimpose the random phase perturbation component, the subharmonic low-frequency phase component, and the low-order aberration tilt phase component, and subtract the spatial average value within the effective area of ​​the superimposed phase screen to remove the piston term, thereby obtaining the mixed turbulence phase screen of this layer. Step S305: Convert the hybrid turbulent phase screen into a complex exponential form of the turbulent phase modulation function.

6. The Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method according to claim 1, characterized in that, Step S4 involves establishing an inclined non-coaxial underwater light propagation geometric model consisting of the transmitter, UOIRS, and receiver. The constructed hybrid turbulence phase screen is embedded into the underwater turbulence propagation path. The evolution of the oblique incident light field from the transmitter to the UOIRS is calculated using the split-step Fourier method. Based on the oblique incident geometry, the oblique incident field in the local coordinate system of the UOIRS is reconstructed, and the complex incident light field on the UOIRS surface, along with its amplitude distribution, intensity distribution, and phase distribution, is output. Step S401: Determine the incident direction and propagation distance from the transmitter to the UOIRS based on the three-dimensional spatial positions of the transmitter, UOIRS, and receiver. Step S402: Construct the initial light field of the emitting end on a transverse plane perpendicular to the incident direction; Step S403: Divide the propagation path from the transmitter to the UOIRS into multiple sub-sections; Step S404: In each step interval, turbulent phase modulation, diffraction propagation and absorption scattering attenuation operations are sequentially applied to the current optical field, wherein the turbulent phase modulation is implemented based on the turbulent phase modulation function of the corresponding layer in step S3; Step S405: Based on the oblique incidence geometry, map the light field reaching UOIRS onto the local coordinate plane of UOIRS, and add the optical path phase caused by oblique incidence to reconstruct the complex incident light field in the local coordinate system of UOIRS.

7. The Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method according to claim 6, characterized in that, Step S5 involves performing phase unwrapping processing on the incident phase distribution of the complex incident light field to establish a phase-matching constraint that enables coherent superposition of the reflected light from each array element at the receiving end, thereby obtaining the ideal reflection phase. The ideal reflection phase is then decomposed into a geometric reflection phase and a turbulence-compensated phase, including: Step S501: Extract the phase distribution of the complex incident light field and perform two-dimensional phase unwrapping processing to obtain the unwrapped phase; Step S502: Using the center of UOIRS as a reference, obtain the referenced incident unwrapped phase; Step S503: Based on the propagation distance of each UOIRS array element to the receiver, establish the phase matching condition that enables the reflected light from each array element to achieve coherent superposition at the receiver, and solve for the ideal reflection phase. Step S504: Decompose the incident unwrapping phase into the ideal geometric incident phase and the turbulent additional phase, and substitute the ideal reflection phase to obtain the geometric reflection phase and the turbulent compensation phase.

8. The Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method according to claim 7, characterized in that, Step S6 involves constructing a Physically Guided Environment-Aware Attention U-Net network. Using the UOIRS complex incident light field, geometric reflection phase, turbulence compensation phase, and marine environmental parameters as inputs, spatial variation compensation weights and residual phase corrections are generated. Based on these spatial variation compensation weights and residual phase corrections, the PEAU adaptive reflection phase is constructed, including: Step S601: The amplitude distribution, phase sine, phase cosine, turbulence-added phase, and geometric reflection phase of the complex incident light field constitute a two-dimensional spatial feature input; Step S602: Use marine environmental parameters to construct a one-dimensional environmental feature input, and map it into environmental features through a feature embedding layer; Step S603: Construct a two-dimensional prior map reflecting the spatial distribution of turbulent distortion based on the phase information of the complex incident light field; Step S604: Input the two-dimensional spatial features, the environmental features, and the two-dimensional prior map into the physical-guided environment-aware attention U-Net network. The network extracts multi-scale spatial features through an encoder and performs adaptive weighted fusion of the encoded features based on the environmental features and the two-dimensional prior map through an attention gating mechanism. Step S605: The network decoder compensates for the weights and residual phase corrections based on the spatial changes in the weighted fused feature output. Step S606: Generate the PEAU adaptive reflection phase based on the spatial variation compensation weight, residual phase correction amount, geometric reflection phase and turbulence compensation phase.

9. The Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction method according to claim 8, characterized in that, Step S7, based on the preset number of phase quantization bits for the UOIRS array elements, discretizes and quantizes the continuously distributed adaptive transmission phase of the PEAU into discrete phases of finite levels. Combining the finite reflection aperture function and reflection efficiency coefficient of the UOIRS, the UOIRS reflection function is constructed and loaded onto each array element, including: Step S701: Perform discrete quantization on the PEAU adaptive reflection phase to obtain the quantized reflection phase; Step S702: Combining the finite aperture function and reflection efficiency coefficient of UOIRS, the quantized reflection phase is constructed into a UOIRS reflection function; Step S703: Multiply the complex incident light field with the UOIRS reflection function point by point to generate the reflected light field after UOIRS modulation.

10. An Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction system, wherein the system is implemented using the Argo data-driven UOIRS underwater optical adaptive wavefront reconstruction described in any one of claims 1-9, characterized in that, The system includes: The Argo data preprocessing module is suitable for acquiring Argo buoy profile datasets of target sea areas, performing standardized preprocessing on the buoy profile datasets, and outputting continuous temperature profiles, continuous salinity profiles, and corresponding vertical gradients under a unified depth grid. The depth-adaptive turbulence spectrum construction module is suitable for constructing a depth-adaptive improved Nikishov seawater refractive index fluctuation power spectrum based on the vertical gradient corresponding to the continuous temperature and salinity profile. The hybrid phase screen generation module is suitable for dividing the underwater light propagation path into multiple thin turbulent propagation layers. For each layer, the module retrieves the corresponding turbulent power spectrum of the depth-adaptive improved Nikishov seawater refractive index fluctuation spectrum at the water depth represented by that layer, generates the hybrid turbulent phase screen corresponding to that layer, and synchronously outputs the corresponding turbulent phase modulation function of that layer. The complex incident light field generation module on the UOIRS surface is suitable for establishing an inclined non-coaxial underwater light propagation geometric model consisting of the transmitter, UOIRS and receiver. The constructed hybrid turbulence phase screen is embedded in the underwater turbulence propagation path. The evolution of the oblique incident light field from the transmitter to the UOIRS is calculated using the step-by-step Fourier method. Based on the oblique incident geometry, the oblique incident field in the local coordinate system of the UOIRS is reconstructed, and the complex incident light field on the UOIRS surface is output. The UOIRS phase matching and decomposition module is suitable for performing phase unwrapping processing on the incident phase distribution of the complex incident light field, establishing a phase matching constraint that enables the reflected light of each array element to achieve coherent superposition at the receiving end, inversely solving to obtain the ideal reflection phase, and decomposing the ideal reflection phase into a geometric reflection phase and a turbulence compensation phase. The PEAU adaptive reflection phase construction module is suitable for constructing a physically guided environmental awareness attention U-Net network. It takes the UOIRS complex incident light field, geometric reflection phase, turbulence compensation phase and marine environmental parameters as inputs to generate spatial variation compensation weights and residual phase corrections. Based on the spatial variation compensation weights and residual phase corrections, the PEAU adaptive reflection phase is constructed. The UOIRS reflection function construction module is suitable for discretizing and quantizing the continuously distributed PEAU adaptive reflection phase into a discrete phase with a finite level according to the preset phase quantization bit number of the UOIRS array element, and constructing the UOIRS reflection function and loading it onto each array element by combining the finite reflection aperture function and reflection efficiency coefficient of UOIRS.

Images