71results about How to "Improve inversion accuracy" patented technology

The invention relates to the technical field of remote sensing data processing and earth surface parameter inversion, and discloses an earth surface matrix parameter hyperspectral data inversion method and system based on machine learning. Comprising the following steps: constructing a multi-source heterogeneous hyperspectral data set; performing feature screening on the preprocessed hyperspectral data set based on an adaptive band selection algorithm, constructing a dynamic weight matrix by calculating mutual information entropy and inter-class distance measurement between spectral bands to realize intelligent screening of key feature bands, and combining spectral derivative conversion and spectral index calculation to generate an enhanced feature vector; and a multi-task transfer learning neural network model is constructed, and an output layer realizes multi-parameter collaborative inversion based on a multi-task learning architecture. And performing preprocessing and feature enhancement operation which is the same as that of the training data on the hyperspectral image data of the to-be-inverted region, inputting the trained neural network model, and outputting a surface matrix parameter inversion result.

The invention provides a mobile terminal skin parameter inversion method based on spectral information, and the method comprises the following steps: constructing an optical model according to the characteristics of a skin layered structure and a spectral response mechanism of three skin parameters: melanin content, hemoglobin concentration and dermis thickness, and building a skin parameter and reflection spectrum database based on the model; based on the database, establishing a skin multi-parameter inversion MLP neural network structure adaptive to the collection characteristics of a mobile phone end sensor, and screening an optimal four-waveband channel through multi-waveband combination traversal; the mobile phone spectrum sensor is subjected to wave band optimization, data acquisition and skin parameter inversion. According to the method, through efficient decoupling of a nonlinear coupling relation among multiple skin parameters by waveband optimization and an MLP model, and targeted design of four optimal visible light narrow-band channels, synchronous quantitative detection of melanin, hemoglobin and dermis thickness is realized, core physiological parameters from epidermis to dermis superficial layer are covered, information coverage is comprehensive, and specificity is high.

This invention discloses a method for inverting sea surface wind speed from GNSS reflected signals based on mask perception and image-based Mamba-CNN. The method includes: performing spatiotemporal matching and interpolation processing on multi-source spaceborne GNSS-R observation data and ERA5 reanalysis data; gridding the processed observation parameters according to geographical location to obtain a multi-channel grid image and acquiring a binarized occupancy mask; inputting the multi-channel grid image and the binarized occupancy mask into a mask perception Mamba-CNN model; calibrating the grid features through a mask conditional encoder and a feature linear modulation layer, and performing spatial modeling using feature extraction blocks combining convolutional and Mamba state space paths to obtain the wind speed inversion result; aggregating the extracted features using center-perception pooling, and outputting the final sea surface wind speed inversion image through a regression network. This invention can significantly reduce computational complexity, effectively reconstruct the continuous wind field structure, and improve the inversion accuracy and robustness in high-wind-speed areas.

This invention relates to a multi-temporal radiation signal inversion method based on Conditional Transformer, belonging to the field of satellite remote sensing data processing / spatial information science and technology. The method includes: acquiring meteorological time-series data of the target area to establish a dataset; designing a Conditional Transformer deep learning network; combining the target time-series data and meteorological data encoded by time and meteorological code generators with the output of the Encoder layer, and inputting this combination into the Decoder layer to train the Conditional Transformer deep learning network to obtain a surface radiation signal inversion model; inverting surface radiation brightness temperature data and evaluating the inversion accuracy; efficiently capturing long-distance spatiotemporal dependencies of radiation signals; flexibly inverting radiation changes according to the target time and meteorological conditions; removing nonlinear interference from the atmosphere on radiation; and more accurately inverting the true radiation characteristics of the surface.

This disclosure provides a lidar wind field inversion method based on a filtered velocity-volume processing algorithm, applicable to the field of lidar technology. The method includes acquiring basic observation data for each voxel in the lidar wind field inversion. This basic data includes the signal-to-noise ratio (SNR) at each observation point within the voxel, used to calculate the average SNR within the voxel. When the average SNR is greater than or equal to a preset SNR switching threshold, the three-dimensional wind vector within the voxel is inverted based on the velocity-volume processing algorithm; otherwise, the three-dimensional wind vector is inverted based on a preset filtered velocity-volume processing algorithm. In this way, the velocity-volume processing algorithm can be used to invert the three-dimensional wind vector from near-field wind field data with a high SNR to improve computational efficiency. In the far-field, when the SNR drops to the preset SNR switching threshold, the three-dimensional wind vector is inverted based on the preset filtered velocity-volume processing algorithm, thereby improving the accuracy of lidar wind field inversion.

The present application relates to the technical field of ecological protection, and more particularly to a forest water conservation capacity remote sensing inversion method suitable for Nanling cloudy and rainy areas, aiming at the problem that the existing inversion method cannot overcome the cloud cover and terrain influence, and the precision of forest water conservation capacity is low, the present application obtains optical and microwave remote sensing data, pre-processes the optical and microwave remote sensing data, extracts corresponding optical characteristics and microwave characteristics, weights and fuses the optical characteristics and microwave characteristics, calculates the forest water conservation amount by using a water conservation amount physical model, and generates a Nanling forest water conservation capacity spatial distribution map. The method overcomes the errors caused by weather, terrain shadow and distortion, constitutes a full-weather, high-precision, Nanling special environment-adapted forest water conservation capacity remote sensing inversion system, and realizes accurate quantification of the water conservation capacity of the Nanling special forest ecosystem.

The water-sand multi-element acoustic online monitoring system and method relate to the technical fields of hydrological monitoring and acoustic measurement. The application is to solve the problems that the traditional single-frequency acoustic monitoring method cannot distinguish the echo intensity changes caused by concentration changes or particle distribution, and it is difficult to realize high-precision flow rate and concentration synchronous monitoring in the same system. The application amplifies and filters the reverberation signal received by the acoustic transducer and extracts the echo envelope. The echo envelope data is calculated by inversion to obtain the water-sand concentration and particle size information. The application has the ability of synchronous measurement of flow rate and concentration, and uses Doppler shift technology to solve the flow rate profile, which provides key data support for sediment transport flux calculation.

The application relates to the technical field of microwave remote sensing, and proposes an adaptive wet troposphere path delay inversion method and system fusing wind speed overlapping partition modeling, multi-scale collaborative sample balancing and model soft fusion. The method comprises the following steps: acquiring multi-source data and preprocessing the multi-source data to obtain to-be-inverted samples; according to a preset threshold, the full-range wind speed of the to-be-inverted samples is divided into a plurality of wind speed subintervals with overlapping transition zones; the to-be-inverted samples are input into trained inversion submodels, the output results of the inversion submodels are weighted and summed according to final fusion weights, and finally, wet troposphere path delay inversion results are obtained; when abnormal wind speed or invalid submodel prediction output is detected, the output result of a pre-trained global unified model is taken as the final inversion result. While guaranteeing overall inversion precision and stability, the application can significantly suppress systematic deviation under sea conditions of extremely low and extremely high wind speed.

This invention discloses a method and system for synchronous inversion of multiple atmospheric parameters, belonging to the field of atmospheric laser remote sensing technology. The probe light and the reference light are obtained by splitting laser pulses; the probe light is expanded, reflected, and then perpendicularly incident into the atmosphere, where it is scattered by the atmosphere to obtain the echo light. That is, all parameters originate from the same measurement and the same spectrum, resulting in perfect spatiotemporal consistency of the data, efficient measurement methods, and accurate measurement results. The temperature and pressure parameters in the atmosphere are calculated using the normalized scattering spectrum, and then aerosol and wind speed parameters are obtained, achieving true synchronous inversion of the four key parameters of temperature, pressure, wind speed, and aerosols for the first time. Furthermore, in applications such as meteorological forecasting, achieving synchronous measurement of atmospheric temperature, pressure, wind speed, and aerosols can save costs and improve efficiency, which has significant practical implications.

This invention discloses an integrated inversion method for cloud macro- and micro-parameters in the interdisciplinary field of atmospheric remote sensing and deep learning. The method includes: inputting multimodal input data into a dual-branch encoder to extract visual multi-scale features and meteorological physical background features; progressively injecting meteorological physical background features into the visual multi-scale features to achieve cross-scale fusion and obtain physically guided decoding features; based on the decoding features, generating spatial gating weights by constructing a cloud area auxiliary map, followed by element-wise gating modulation and feature refinement to obtain spatially focused features; inputting the spatially focused features into a multi-task prediction head to obtain the integrated inversion results of four cloud macro- and micro-parameters: cloud mask probability map, cloud top height, cloud optical thickness, and effective radius of cloud particles. This invention can solve the problems of difficulty in effectively fusing high-resolution satellite visual features with low-resolution atmospheric background fields and insufficient physical consistency caused by independent inversion of multiple cloud macro- and micro-parameters.

The application discloses a trace gas supersampling method and system based on TROPOMI satellite data, and relates to the fields of atmospheric environmental science and remote sensing technology.The method comprises the following steps: obtaining effective remote sensing observation data by preprocessing remote sensing observation data of a target research region acquired by a troposphere detector TROPOMI L2, and obtaining corrected effective remote sensing observation data by performing multiplication factor correction processing on the effective remote sensing observation data; constructing a target grid system according to the longitude and latitude of the target research region and a target grid resolution, and establishing spatial indexes for each grid unit in the target grid system by using R-trees; and resampling the corrected effective remote sensing observation data into the target grid system by using a supersampling method to obtain the concentration value of trace gas in each grid unit in the target grid system.

The invention discloses an ear plate local stress high spatial resolution monitoring method and system based on ultra-weak grating spectral shape demodulation. The system comprises an optical fiber sensing array, a light source module, a detection module, a data acquisition and physical information neural network inversion module and a display alarm module. The optical fiber sensing array is composed of N ultra-weak gratings arranged in a stress concentration area of the ear plate and is used for sensing local strain; the light source and detection module adopts a time division and wavelength division multiplexing strategy to realize synchronous monitoring of multiple ear plates; the data acquisition and inversion module acquires stress distribution along the length direction of the ultra-weak grating based on a physical information neural network inversion model; and the display alarm module realizes local stress threshold alarm of the ear plate. According to the invention, high-precision and real-time monitoring of local stress of metal connection nodes such as ear plates can be realized, the method is suitable for long-term online health monitoring of large-scale structure connection areas such as bridge inhaul cables, steel trussed beams and cable bent towers, and technical support is provided for structural fatigue early warning and safety assessment.

This invention discloses a method for identifying the deterioration of subgrades of long-service highways based on a shear modulus inversion model. It utilizes various sensors embedded in the subgrade to acquire multi-source monitoring data of the target road segment, including environmental parameters, mechanical response parameters, physical state parameters, and spatial location information, and performs preprocessing. By constructing hysteresis features and removing recursive features, an in-situ subgrade shear modulus inversion model is established, transforming the multi-source monitoring data into a shear modulus distribution field that considers the temporal changes in the subgrade's state and includes spatial information. Based on the inverted spatial distribution field of the shear modulus, spatial variability indices such as the coefficient of variation and Moran's index are calculated, and the subgrade deterioration state is identified based on its evolution trend, outputting the deterioration identification result. This invention provides a scientific basis for maintenance decisions.

The present application belongs to the field of laser radar signal processing, and provides a method and system for retrieving vertical visibility based on Mie scattering laser radar, the method comprising the following steps: S1, acquiring signals; S2, signal preprocessing, sequentially comprising background noise correction, overlap factor correction, distance square correction, denoising processing and normalization; S3, vertical visibility retrieval, comprising the construction, training and use of a DBO-BP optimized neural network model. The system comprises a Mie scattering laser radar, a data preprocessing module and a DBO-BP optimized neural network model module. When retrieving vertical visibility, the present application takes into account the physical characteristics and data characteristics of the original echo signal, significantly enhances the anti-noise performance through multiple denoising when preprocessing the signal, and balances the efficiency and accuracy by adopting a hierarchical iteration strategy when training the model.

This invention relates to the field of remote sensing water quality identification technology, and in particular to a method for identifying black and odorous water bodies based on a temporal convolutional network. By constructing a black and odorous water body identification model comprising a cascaded temporal convolutional network residual module, a channel attention module, a temporal attention module, a global average pooling layer, and a regressor module, and replacing the original output layer of the temporal convolutional network with the regressor module, a mapping from remote sensing image data to continuous water quality parameter predictions is achieved. This solves the technical problem of limited accuracy in identifying black and odorous water bodies by relying solely on general spectral features in existing technologies. Furthermore, by performing addition, subtraction, multiplication, and division operations on the original spectral bands to generate combined bands, and using the Pearson correlation coefficient method to select the original and combined bands with the highest correlation, a sample dataset is constructed by combining this with ground-based measured data. This further expands the sample size based on existing samples, significantly improving the model's learning efficiency.

This invention relates to the field of coherent wind lidar technology, and discloses a passive solid-state beam scanning system based on wavelength tuning and dispersive grating coupling. A tunable laser unit outputs multiple narrow-linewidth lasers of different wavelengths; a beam-splitting modulation unit divides them into emitted light and local oscillator light, and pulse-modulates the emitted light; a dispersive grating scanning unit forms scanning beams with different exit directions from the emitted light of different wavelengths; a wavelength monitoring unit acquires the actual wavelength of the emitted light, and a grating state monitoring unit acquires the grating state parameters; a joint calibration and compensation unit determines the actual exit angle based on the actual wavelength of the emitted light and the grating state parameters, and determines the Doppler velocity conversion coefficient based on the actual wavelength of the emitted light; a coherent receiving and solving unit calculates the radial wind speed, and a wind parameter inversion unit inverts the wind parameter information. This invention can jointly correct the scanning direction and velocity conversion, improving the accuracy of wind parameter information inversion and long-term wind measurement stability.

The invention belongs to the technical field of electromagnetic inversion, and discloses a transient electromagnetic artificial intelligence inversion method, system and device based on physical guidance, and a medium, and the method comprises the steps: obtaining electromagnetic response signal data; inputting the electromagnetic response signal data into a transient electromagnetic inversion model for inversion prediction to obtain underground resistivity structure data; wherein the transient electromagnetic inversion model is constructed based on a convolutional neural network and a long short-term memory network and comprises a spatial feature extraction module, a sequential relation modeling module, an attention enhancement module and a resistivity reconstruction module which are connected in sequence, and a pre-training forward modeling network is further embedded in the transient electromagnetic inversion model. The pre-trained forward modeling network is used for adding physical constraints to the transient electromagnetic inversion model. According to the technical scheme of the invention, the calculation efficiency can be improved, and the anti-noise performance of the model and the reliability of the inversion result are enhanced.

This invention relates to the field of quantitative remote sensing parameter inversion technology, specifically to a deep learning inversion method for passive microwave cloud-to-surface temperature constrained by physical mechanisms. The method includes the following steps: S1: acquiring model input data and preprocessing it; S2: constructing a deep learning inversion model framework constrained by physical mechanisms; S3: training the deep learning inversion model and performing cloud-to-surface temperature inversion. This invention fully combines the advantages of physical model methods in improving the physical interpretability and universality of surface parameters with the advantages of neural network algorithms in handling complex and nonlinear problems. It can solve the problem of cloud-to-surface microwave surface temperature estimation, which is difficult to achieve with high inversion accuracy, clear physical meaning, and strong generalization ability using traditional methods.

This invention is a method for inverting azimuth difference seismic data in inclined fractured rocks. Previous AVAZ inversion methods for fracture parameters were mostly based on the HTI medium assumption, with insufficient consideration for fracture dip angle. The fixed form of the inversion constraint terms made it difficult to adapt to changes in parameter statistical characteristics, and the regularization parameters were all manually adjusted, affecting inversion accuracy and efficiency. This invention establishes an approximate formula for the reflection coefficient of the TTI medium, characterized by the P-wave and S-wave velocities, density, and fracture elastic parameters of the background medium. It utilizes pre-stack azimuth difference data to perform AVAZ inversion of fracture parameters. Dynamic sparsity constraints are introduced, dynamically adjusting the sparsity of the constraints based on data characteristics, and a covariance matrix is ​​introduced to consider the statistical regularity between parameters. Combining iterative reweighted least squares and adaptive regularization parameter acquisition methods, the dynamic constraint inverse problem is transformed into a weighted... L A 2-norm objective function is used to achieve a stable solution. Data testing shows that this invention can effectively improve the inversion accuracy of crack elastic parameters.

This invention discloses a layered inversion method and system for aerosol lidar based on polarization information. The method includes: acquiring parallel and vertical polarization channel echo signals from a polarized Mie scattering lidar; performing layered identification of aerosols based on the total echo signal to determine the altitude range of upper-air and near-surface aerosol layers; for upper-air aerosol layers, calculating the estimated aerosol depolarization ratio based on the polarization channel echo signals and establishing a quantitative inversion relationship between the volume depolarization ratio and the aerosol backscattering coefficient, thereby achieving the inversion of the upper-air aerosol backscattering coefficient; for near-surface aerosols, inverting the backscattering coefficient using the Fernald method; and stitching the inversion results from different altitude regions according to the layered information to form a backscattering coefficient profile across the entire altitude range. Using this invention, the stability and reliability of upper-air aerosol inversion can be effectively improved.

The invention discloses a rotating electromagnetic field surface corrosion defect inversion system and method and a medium, and belongs to the technical field of corrosion defect quantitative detection, and the system comprises a rotating electromagnetic field excitation module which is used for generating a rotating magnetic field excitation signal to act on the surface of a detected metal component; the three-axis magnetic field acquisition module is linked with the rotary electromagnetic field excitation module and drives a three-axis magnetic field sensing probe to complete scanning of a to-be-detected area; the feature tensor construction module is used for receiving the three-axis magnetic field matrix output by the three-axis magnetic field acquisition module; the loss function construction module is used for constructing a mixed loss function; the inversion neural network module is used for outputting a three-dimensional shape prediction result of the corrosion defect; and the model deployment module deploys the trained inversion neural network optimal model to an actual detection system. By constructing the multi-dimensional magnetic field coupling feature tensor and combining the mixed loss function of the double gradient constraint, the problems of low inversion precision and poor anti-interference capability of the three-dimensional morphology of the corrosion defect under the complex working condition are solved.

This invention relates to the field of foundation pit construction safety monitoring technology, and provides a method and system for foundation pit monitoring analysis and geotechnical parameter inversion. The method includes: building a finite element model of the foundation pit to generate a batch of simulation monitoring curve datasets and binding labels; preprocessing the simulation monitoring curve data and field data respectively, and performing logarithmic transformation on the permeability coefficient labels of soft clay and confined aquifers; constructing and training a two-level integrated model composed of a classification neural network and multiple sets of regression subnetworks; inputting field data into the trained model for two-level inference parameter inversion; performing an inverse exponential transformation on the permeability coefficient to restore the true value and performing risk classification and early warning. This invention can replace measured samples with simulation data, improve inversion accuracy through a two-level integrated architecture, and realize intelligent monitoring and parameter inversion of complex strata.

The invention relates to the technical field of body armor testing, in particular to a body armor damage positioning method and system based on a flexible sensor network, and the method comprises the steps: collecting a multi-mode impact signal generated by an impact event through a partition heterogeneous flexible sensor network, and synchronously collecting the posture data of a wearer; carrying out adaptive anti-interference preprocessing on the multi-mode impact signal, and dynamically adjusting a filtering parameter according to the attitude data to obtain a pure impact signal; the signal receiving time difference of the sensing nodes is extracted, dynamic medium coefficient calibration is carried out in combination with sensing network deformation characteristics, and impact point plane coordinates are solved through an optimization algorithm; extracting impact energy related parameters to divide energy levels, inverting the penetration depth of an impact object through an energy attenuation model, and forming damage three-dimensional coordinates with impact point plane coordinates; and dynamically matching the damage grading threshold values to obtain a quantitative damage result. According to the scheme, the damage positioning and evaluation precision is greatly improved through an accurate adaptation mechanism of the special medium characteristics of the body armor and the dynamic deformation.

The invention is suitable for the technical field of ecological environment, and provides an urban carbon emission inversion method based on WRF-FLEXPART, and the method comprises the steps: simulating the meteorological data of a research region in a preset time period through a WRF model, and obtaining the simulated meteorological data; inputting the simulated meteorological data into a FLEXPART model, and simulating the dissipation trajectory of the gas particles by using the FLEXPART model to obtain the contribution degree of each grid unit in the research area to the carbon dioxide concentration at the urban site within a preset time period; respectively acquiring the carbon dioxide concentrations of the urban site and the background site of the research area in a preset time period; and based on the contribution degree corresponding to each grid unit and the carbon dioxide concentration of the city site and the background site within the preset time period, obtaining the carbon dioxide emission of each grid unit. According to the method, the timeliness and precision of inversion can be improved while the workload is reduced.

This invention discloses an adaptive variable incident angle ocean salinity detection method and apparatus. The method includes: dividing the global ocean area into regions; setting classification standards for sea surface temperature, sea surface salinity, and sea surface wind field; mapping the sea surface temperature data, sea surface salinity data, and sea surface wind field data corresponding to each region based on the classification standards to obtain sea state vectors for each region; and determining the incident angle sequence of the two-dimensional synthetic aperture radiometer (SAM) carried on the satellite when the satellite flies over the region where the SAM incident angle sequence is to be generated, based on the sea state vector of the region, with each region corresponding to one incident angle sequence. This method can increase the proportion of a specific incident angle range in a specific region, thereby addressing the shortcomings of obtaining high-precision salinity for specific regions and improving the accuracy of global ocean salinity detection.

The application discloses a static satellite ground surface emissivity cooperative inversion method based on background field reconstruction and belongs to the technical field of satellite remote sensing information processing, which comprises the following steps: constructing a background field of spatial structure cooperation constraint of ground surface temperature and atmospheric temperature, adopting a three-dimensional variation assimilation cost function to assimilate the measured ground surface temperature of a ground meteorological observation station, obtaining a ground surface temperature analysis field as a ground surface temperature initial guess field of a one-dimensional variation inversion cost function to invert the ground surface emissivity; generating a ground surface emissivity background field from the ground surface emissivity in a ground surface element historical data set, constructing a ground surface item background error covariance matrix capable of representing the coupling relationship between the ground surface emissivity and the ground surface temperature based on the ground surface element historical data set, and further constructing a one-dimensional variation inversion cost function; and combining the satellite observation brightness temperature of a clear sky observation pixel observed by an infrared window channel of a geostationary orbit radiometer, so that the inversion of the ground surface emissivity can be realized, and thus the ground surface emissivity inversion level is improved.

The present application belongs to the field of artificial intelligence geoscience application and dense reservoir evaluation technology, and particularly to a dense reservoir structure porosity inversion method based on rock physical prior embedding. The method comprises the following steps: S1: obtaining conventional logging data and calibration data; S2: logging data preprocessing; S3: constructing rock physical prior features; S4: constructing a rock physical prior embedding input matrix; S5: constructing an integrated coupling model; S6: extracting bidirectional sequence features in the well depth direction; S7: performing nonlinear inversion; S8: constructing a rock physical prior constraint loss function; S9: performing joint global optimization; and S10: outputting intelligent inversion results of structure porosity. The continuous curves of free porosity, micro-microporosity and clay porosity output by the present application can directly serve reservoir effectiveness evaluation, movable fluid identification, reserve calculation and development interval optimization, and provide reliable parameter support for oil and gas exploration and development.

The application relates to the field of exploration geophysical seismic fluid identification, in particular to a frequency-variable solid-liquid decoupling viscoelastic fluid factor prestack seismic inversion method. The method comprises the following steps: step 1, carrying out equal-interval frequency division processing on seismic data; step 2, carrying out frequency division elastic impedance inversion by using the frequency division seismic data in step 1; step 3, frequency-variable solid-liquid decoupling fluid factor inversion equation construction; step 4, carrying out frequency-variable solid-liquid decoupling fluid factor inversion equation coefficient inversion for each prestack angle stack and sampling point; and step 5, frequency-variable solid-liquid decoupling fluid factor inversion. The method improves the fluid factor inversion precision on the basis of conventional solid-liquid decoupling fluid factor inversion by establishing a reference frequency and frequency division frequency calculation process.

A method based on local multi-source main scattering to whole-vehicle scattering characteristics is proposed. This method divides a complex target into multiple target structural regions, acquires high-resolution scattering data for each local region using a multi-frequency, multi-polarization radar system, and establishes a mapping relationship between local and whole-vehicle scattering behavior by combining imaging processing and main scattering source extraction. Furthermore, a dynamic weighting mechanism is introduced to adaptively adjust the influence weight of each region in the whole-vehicle inversion based on its contribution and measurement confidence, thereby improving the stability and robustness of the inversion results. This invention effectively alleviates the feature loss problem caused by limited viewing angle and structural complexity in traditional measurements and is applicable to multiple application scenarios such as aircraft radar scattering modeling, stealth performance evaluation, and electromagnetic characteristic analysis.

The application discloses a kind of face rainfall distribution reconstruction method and device based on space-time graph neural network, wherein the method first constructs space-time graph structure model, the complex spatial correlation between links is established by the adjacency matrix construction method of multi-factor fusion, and the graph structure data object containing space-time dimension is constructed;Further, hierarchical space-time graph neural network is used for feature extraction and fusion, and the space-time depth feature representation of the fused rainfall field is obtained;In the model optimization process, the microwave rain attenuation physical model is introduced as a constraint condition simultaneously, and the physical consistency loss function is used to ensure that the inversion result conforms to the electromagnetic wave propagation law;Based on the grid rainfall intensity and its uncertainty evaluation output by the network, high-resolution surface rainfall distribution products are generated through spatial interpolation and quality control;Finally, a real-time inference system and feedback mechanism are established to dynamically update and output the rainfall field, providing an intelligent technical solution from "line measurement" to "surface reconstruction" for meteorological and hydrological monitoring.