79results about How to "Restore accurately" patented technology

The invention relates to the technical field of magnetic suspension compressor assembly, and particularly discloses a magnetic pollution source prevention and control optimization method in a magnetic suspension compressor assembly process, which comprises the following steps: in a pretreatment stage of magnetic suspension compressor assembly, dynamically optimizing thermal cycle parameters in a low-temperature environment by adopting a self-adaptive genetic algorithm; ferromagnetic impurities hidden in microdefects of the compressor component are accurately induced to be released in advance; carrying out feature extraction and pattern recognition on the released impurity signals in combination with wavelet packet transformation and a neural network algorithm; according to the method, a low-temperature thermal cycle process optimized by a self-adaptive genetic algorithm is constructed in the pre-assembly treatment stage, ferromagnetic impurities hidden in microcracks are accurately induced to be released in advance, and accurate removal of the impurities before entering the assembly link is realized by combining wavelet packet transformation and targeted pulse demagnetization driven by a neural network; the limitation that impurities in the microdefects cannot be touched by a traditional cleaning and demagnetizing method is broken through, and the hidden danger of magnetic pollution which cannot be found by initial detection is eliminated from the source.

This invention discloses a concrete testing device for road and bridge construction, relating to the field of concrete testing. It includes a sample holder and a movable pressure plate. Vertical plates are fixed to both sides of the upper surface of the sample holder, and top plates are fixed to the top of each of the two vertical plates. A limiting groove for placing concrete blocks is formed at the center of the upper surface of the sample holder. The pressure plate is located above the limiting groove, and a groove is formed at the center of the upper surface of the pressure plate, in which a liquid storage cylinder is fixed. Liquid permeation holes are formed at the groove and at the bottom of the liquid storage cylinder on the lower surface of the pressure plate. A pressure application component and two transmission components are provided above the liquid storage cylinder. The two transmission components are symmetrically arranged on the pressure application component. Each of the two rope-winding rollers has an inner sprocket tooth on one corresponding spoke. This invention achieves simultaneous testing of concrete compressive strength and impermeability, shortens the testing cycle, and restores the stress environment of road and bridge concrete during actual service, improving the comprehensiveness and reliability of the test data.

The application discloses a kind of based on deep neural network's tire ground pressure distribution form prediction method, method includes: generating simulation sample;Calibration is carried out to simulation sample, and calibration sample is obtained;Performing physical information-driven data enhancement operation, integration generates multi-working condition sample library;Deep neural network model is built;Deep neural network model is iteratively trained and model weight solidification is completed, and prediction model is obtained;It is predicted;The deep neural network of the application is constructed by encoder-decoder architecture, and the tire ground pressure distribution calculation that traditional finite element simulation needs several hours to complete is compressed to millisecond order, and the technical bottleneck of time-consuming physical simulation is broken through.In the absence of any physical sample, only by inputting structure design parameters and working condition parameters, the pressure distribution form covered in full working condition can be obtained in real time, supporting the rapid screening and optimization iteration of hundreds of tire structure schemes in a single day, significantly shortening the tire development cycle.

This invention relates to the technical field of simulation analysis methods for downhole tools in oil and gas fields, and discloses a high-precision dynamic simulation analysis method for full-size drilling and milling tools based on an adaptive FEM-SPH coupled algorithm. The method includes three processes: model establishment, model solving, and performance evaluation. The first process is model establishment, which involves obtaining the geometric model of the drilling and milling tool through model reconstruction technology, and establishing a high-precision dynamic simulation analysis model for the full-size drilling and milling tool based on a wear prediction model and the adaptive SPH-FEM coupled algorithm. The second process is model solving, which involves solving the model after successively changing the drilling pressure and rotational speed, and batch processing the simulation results using Python. The third process is performance evaluation, which evaluates the performance of the drilling and milling tool from three aspects: efficiency, safety, and lifespan. This invention improves simulation accuracy from three aspects: geometric modeling, wear simulation, and solution algorithm, and evaluates the performance of the drilling and milling tool from three aspects: efficiency, safety, and lifespan.

The application is suitable for the technical field of data processing, and provides a negative feedback monitoring method applied to media data, which comprises the following steps: step S1: media data collection and fusion: collecting media data and user interaction behavior data from different channels, and constructing a unified user session sequence model to time-correlate multi-modal data within a single user session; step S2: multi-modal negative feedback feature fusion: processing the media data, extracting negative feedback features of text, vision, audio and user behavior modes, and analyzing the correlation between different modal negative feedback features by using a cross-modal correlator. The application effectively avoids the problem of cross-session data fragmentation by limiting the operation range of the user session sequence model to a single continuous access period.

This invention relates to a super-resolution reconstruction method for UAV images of kiwifruit sunburn. Compared with existing technologies, it solves the problems of poor model generalization due to the lack of accurate modeling of the actual physical degradation of UAVs, and the difficulty of feature extraction networks in simultaneously considering global semantics and local high-frequency lesion details, resulting in blurred reconstruction. The invention includes the following steps: acquisition and preprocessing of UAV images of kiwifruit sunburn; generation of multi-scale paired datasets; construction of a super-resolution reconstruction model for kiwifruit sunburn images; training of the super-resolution reconstruction model for kiwifruit sunburn images; acquisition and preprocessing of UAV images; and super-resolution reconstruction of UAV images. This invention significantly eliminates boundary artifacts in image processing by constructing a multi-scale attention and spatial-frequency fusion mechanism, accurately restoring the high-frequency edge texture details of small lesions at different scales, and reducing data acquisition time.

The embodiment of the invention provides a medical knowledge graph construction method and device. In the method, a page in an electronic file of a first medical literature is input into a large model, entity data of entities in the page are obtained through recognition, the recognized entity data are combined into a constructed knowledge graph through semantic similarity matching, the combination comprises the steps that the entities serve as nodes, knowledge points serve as attributes of the nodes, and the attributes of the nodes serve as attributes of the first medical literature. And constructing a typeless connection edge between the entity and the associated entity thereof. And when the entity data corresponding to all the pages of the electronic file are merged into the constructed knowledge graph, obtaining the medical knowledge graph of the first medical literature. Wherein the entity data comprises entity names, entity types, associated entities and knowledge points related to the entities. The knowledge point text is an original sentence of text content in a page or a slightly modified text, and contains complete and rich association relationship information between associated entities. Private data need to be protected in the construction process.

This invention relates to the field of gas pipeline safety monitoring technology, and discloses a method for predicting the risk situation of underground gas pipelines based on neural network 3D modeling. The method includes: collecting multi-source heterogeneous data of the pipeline network and preprocessing it to generate a standardized fusion dataset; extracting spatial and temporal features based on a fusion neural network of CNN and LSTM, and generating a 3D model of the pipeline network by combining GIS coordinates; analyzing the risk factor weights using the Grey-DEMATEL-ISM method, and using the N-K model to calculate multi-factor coupled risks to generate the coupled risk level of each pipe segment; predicting the risk situation based on the 3D model features and coupled risk results, and outputting graded early warning information.

The application discloses a method for supporting multi-terminal touch screen information collection and restoration. A data collection terminal collects touch screen data on different signing terminals. A handwriting data conversion module in a data processing layer pre-processes the touch screen data obtained from the different signing terminals, converts multi-terminal handwriting data into standard handwriting data packets in a screen coordinate system, determines key frames of the touch screen data, fills in the blank between adjacent point positions of the touch screen data by using a tweening animation algorithm to ensure that there is no broken line, calls a Bezier curve algorithm to determine the intermediate frame inserted between the two key frames, so that the handwriting is smooth, and the obtained touch screen data is converted into lines close to the natural handwriting habit of a person. A rendering layer determines the size of a handwriting point according to the coordinates of a current point position on a canvas, real-time pressure data and a predetermined thickness, renders a handwriting track, and generates an electronic signature.

The application provides a nuclear fuel grid micro-oscillation fatigue failure evaluation method, and relates to the technical field of nuclear fuel grids, and comprises the following steps: obtaining an elastic wave signal and a normal clamping force of the nuclear fuel grid, then removing noise information in the signal to extract a pure high-frequency micro-oscillation impact signal, combining the normal clamping force to construct a time-domain load spectrum, calculating fatigue damage according to the load spectrum, and judging whether the grid fails or not. Through effective noise reduction and feature extraction on the elastic wave signal in a strong noise background, the real time-domain characteristics of the micro-oscillation impact can be accurately restored, so that the key time node of the transition of the contact surface from adhesion to sliding can be accurately positioned, the accuracy and reliability of the nuclear fuel grid micro-oscillation fatigue failure evaluation are significantly improved, the traditional offline estimation is realized to the full life cycle online evaluation based on the real dynamic load, and the structural integrity of the nuclear fuel assembly under the long-term service condition is effectively ensured.

The application discloses a yaw determination method and device, computer equipment and a storage medium, the method comprises the following steps: when the target vehicle yaw, obtaining the original trajectory point information, the actual yaw time, the path information of the re-planned path and the time of the re-planned path of the target vehicle; based on the original trajectory point information, determining the actual driving path of the target vehicle; based on the actual yaw time, the time of the re-planned path and the actual driving path, determining the yaw trajectory; based on the yaw trajectory and the planned path before the target vehicle yaw, determining whether the target vehicle is in an error-prone scene; if the target vehicle is in an error-prone scene, determining whether the yaw misjudgment occurs. Not only can the driver's driving path be restored more accurately, but also the original trajectory point can be compatible with the complex scenes such as the road network equidistance deviation and the parallel road, so as to determine whether the current yaw determination of the navigation engine is wrong, thereby improving the accuracy and integrity of the yaw determination of the navigation engine.

The invention belongs to the technical field of high-temperature coke ignition time measurement, and particularly relates to a device and a method for measuring high-temperature coke ignition time delay. The device for measuring the ignition time delay of the high-temperature coke comprises a plane flame burner, an image acquisition unit and a data processing unit, in the plane flame burner, a top preset section of a central powder feeding pipe extends into a hearth; the image acquisition unit is used for acquiring a motion trail diagram that coke particles output by the top preset section are gradually heated to be on fire; and the data processing unit is used for performing brightness data extraction according to the motion trail diagram to obtain ignition time, and taking the time interval from the time when the coke particles are output into the hearth from the top preset section to the ignition time as ignition time delay output of the high-temperature coke. According to the invention, the technical problem that only the ignition delay of normal-temperature pulverized coal can be effectively measured and the high-temperature coke with a certain temperature cannot be accurately measured in the prior art is solved.

The utility model discloses a substitute body for dental implant, the substitute body is made of resin material and comprises a butt joint part, a positioning section and a fixing section, the butt joint part is arranged inside the fixing section, the butt joint part is connected with an abutment, the positioning section is connected with a dental cast body, the positioning section, the connecting part and the fixing section are arranged along the same central axis, and the butt joint part is connected with the positioning section. A butt joint groove is formed in the center of the butt joint part, the butt joint groove comprises a transition section and a clamping groove with the cross section being in a regular polygon shape, arc-shaped protrusions are arranged on the outer surface of the positioning section, the number of the protrusions is the same as that of the sides of the clamping groove, and the positions of the sides of the clamping groove correspond to the positions of the protrusions in a one-to-one mode. The oral cavity replacement device is reasonable in structural design, conditions in the oral cavity of a patient can be restored more sufficiently after the base is installed on the replacement body, and the replacement body can be installed and adjusted more conveniently; a dentist technician accurately restores the condition in the oral cavity of a patient through the replacement body, and the manufactured false tooth is excellent and is more adaptive to the oral cavity of the patient.

The embodiment of the invention discloses a page code generation method and device, a storage medium and a computer program product, and the specific implementation scheme is as follows: inputting a page design drawing into a drawing code model to generate an initial page code; converting the initial page code into a page effect picture; determining difference information between the page design drawing and the page effect drawing; the difference information and the page design drawing are input into a drawing code model to generate a target page code, and the target page code is used for generating a target page corresponding to the page design drawing. Based on the scheme, the page code of the corresponding target page can be automatically generated according to the page design drawing, the development efficiency of the page code is improved, the page code can be optimized according to the difference information of the generated page effect drawing and the original page design drawing, the quality of the page code is improved, and the user experience is improved. And accurate restoration of the page design drawing is realized.

This invention discloses a cable production identification and monitoring system and method based on cable images, belonging to the field of image recognition technology. The system includes: acquiring a sequence of cable surface depth images to construct a grayscale surface map along the cable surface depth direction; using the central ridge of the grayscale surface map as a dynamic axis of symmetry, detecting candidate pixels through multi-scale morphological top-hat transformation to form a symmetrical residual distribution map; combining the eight-neighbor gradient direction field and Hessian matrix curvature to remove false defect points, obtaining an effective image, and inputting it into a pre-trained multi-task neural network to simultaneously output a defect segmentation mask and category label; generating a three-dimensional mesh model with dimensional accuracy through Poisson surface reconstruction; and finally, using the physical coordinates of the defect center to drive a laser marking machine to mark a QR code containing key information, which is then associated and stored in a blockchain database. This application improves the accuracy and efficiency of defect identification, realizing the quantitative characterization and full lifecycle management of defects.

A carbon brush detection device of the present invention comprises a main body, a length detection assembly, a processor and a pressure detection assembly, the length detection assembly comprises a fixed probe rod, a movable probe rod, a displacement sensor and a driving assembly, one end of the fixed probe rod is arranged on the main body, the other end of the fixed probe rod is provided with a top plate, and the movable probe rod is movably arranged on the main body. The displacement sensor is arranged on the fixed feeler lever to detect the moving distance of the movable feeler lever, and the driving assembly is connected with the movable feeler lever to drive the movable feeler lever to move; the processor is electrically connected with the displacement sensor; the pressure detection assembly comprises a tension meter and a fixing band, the tension meter is slidably arranged on the main body, and the fixing band is arranged at the measuring end of the tension meter. According to the carbon brush detection device, the pressed state of the carbon brush in the actual state is accurately restored, deviation of detection data can be effectively avoided, the detection accuracy of the device is improved, meanwhile, in the detection process, multiple detection tools do not need to be used, and use is more convenient.

The invention discloses a full-flow microsecond-level measurement method and system based on data plane micro burst measurement, and the method specifically comprises the steps: carrying out the time-slot flow statistics of a data flow in a data plane, and obtaining a current time-slot flow observation value and a previous time-slot flow observation value; calculating a first-order difference based on the current time slot flow observation value and the previous time slot flow observation value to obtain a flow rate variable quantity; on the basis of the flow rate variable quantity and the historical cumulative variable quantity, lightweight chi-square testing is executed through a logarithmic projection technology, and a burst score is calculated; comparing the burst score with a preset threshold value, and if the burst score exceeds the preset threshold value, generating a burst report and reporting the burst report to a control plane; and receiving burst reports on a control plane, filling data between adjacent burst reports, and reconstructing a continuous flow rate curve of a data stream. According to the method, the contradiction among communication overhead, hardware adaptability and measurement granularity of the existing method is solved, so that high-fidelity and low-overhead microsecond-level telemetering is realized.

The utility model belongs to the technical field of lighting devices, and particularly provides a lighting system which comprises a control module, an environment detection module, a white light emitting module and a golden light emitting module, and the golden light emitting module comprises a yellow light emitting assembly and a red light emitting unit. When the environment detection module detects that the weather condition is good, the control module only lightens the white light emitting module, and when the environment detection module detects that the weather condition is bad, the control module lightens the golden yellow light emitting module and simultaneously closes the white light emitting module or keeps lightening the white light emitting module. Yellow light emitted by the yellow light emitting assembly and red light emitted by the red light emitting unit are mixed to obtain golden yellow light with the color temperature ranging from 1700 K to 2500 K, compared with white light, the golden yellow light is lower in color temperature, longer in wavelength and weaker in scattering in environments such as rain, fog and dust, penetrating power can be improved, the effective lighting distance can be increased, and therefore driving safety is improved.

The application proposes a large-scale MIMO system channel graph construction method based on a latent space diffusion model (LDM). The method uses the spatial consistency of the large-scale MIMO system channel to learn the conditional probability distribution of the statistical channel state information (sCSI) about the user terminal position to construct the channel graph. An improved variational autoencoder (VAE) is used to learn the low-dimensional representation of high-dimensional sCSI, and the LDM is improved to improve the generation accuracy of the low-dimensional representation of sCSI, and then a channel graph containing position and sCSI low-dimensional representation data is constructed. The storage and transmission cost of the channel graph is greatly reduced compared with directly using high-dimensional sCSI, and the sCSI low-dimensional representation in the channel graph can be accurately restored to high-dimensional sCSI using VAE, and the generation process does not depend on the pilot signal, which greatly saves the pilot overhead while accurately generating sCSI.

This application relates to the field of Internet of Things (IoT) technology and discloses a method, apparatus, electronic device, and storage medium for IoT device access. The method includes: extracting message features corresponding to message data reported by the device to be accessed; calculating the confidence level between the message features and protocol fingerprints in a protocol fingerprint database; the protocol fingerprint database includes protocol fingerprints corresponding to multiple communication protocols, with each communication protocol corresponding to one protocol fingerprint; determining whether the maximum confidence level is greater than a preset threshold; if the maximum confidence level is greater than the preset threshold, accessing the device to be accessed based on the communication protocol corresponding to the maximum confidence level; if the maximum confidence level is less than or equal to the preset threshold, creating a corresponding parser instance based on the message features, and accessing the device to be accessed based on the parser instance. This application can improve the flexibility and efficiency of device access.

This application relates to the field of target detection technology and provides a method and apparatus for determining roadside base station installation information. In this method, a high-precision map is used as the basic simulation environment, combined with sensor simulation techniques, to accurately reconstruct the perception results of roadside base stations within the target area. Furthermore, considering that pedestrian data is typically limited during simulation, potentially leading to the omission of blind spots and affecting the installation layout of roadside base stations, a pedestrian model is incorporated to update the blind spot information. The resulting base station installation locations further enhance the perception capabilities of the roadside base stations.

This application provides a video privacy protection method, a video decryption method, and related apparatus. The video encryption device uses an initialization vector and an encryption key to encrypt video frames containing privacy information in the original video, thereby achieving privacy protection. During the compression encoding of the privacy-protected video, the video encryption device uses a preset password to encrypt the initialization vector and / or the encryption key, and writes the encrypted parameter data into the video encoded data. On the one hand, for unauthorized attackers who do not know the preset password, they cannot accurately reconstruct the original video, thus further ensuring information security. On the other hand, for authorized users with the preset password, they can utilize the complete reversibility of encryption and decryption to accurately decrypt the lossless compressed encoded data, accurately reconstructing the original video, achieving reversible video privacy protection.

The application is an elongated marine structure deformation monitoring method based on a moving boundary virtual extension, and belongs to the technical field of marine engineering structure monitoring based on computer data processing; aiming at the problem of curvature extrapolation instability caused by dynamic disturbance of the structure end, the application constructs a virtual extension boundary at the structure end, establishes a parameterized constraint model with the extension length and the curvature adjustment coefficient as variables, and determines the optimal virtual boundary geometric form with the minimum reconstruction error minimization as the target; the entity and the virtual boundary are spliced into a global analysis domain, and the three-dimensional deformation curve of the central axis is reconstructed based on sparse strain data. The application converts the structure end moving boundary problem into an optimal curvature constraint, effectively suppresses the curvature mutation and error accumulation, and significantly improves the accuracy and stability of the structure deformation reconstruction under the weak constraint condition of the end.

The present application relates to the technical field of finite element modeling, in particular to a kind of chest impactor or dummy two-dimensional displacement sensor finite element modeling method.The present application includes: respectively carrying out geometric modeling and meshing for slide rail sleeve, support and other components;For slide rail sleeve, set up elastic-plastic material, for support, set up rigid material;Arrange node in measuring point, the most front end of slide rail sleeve and support interior, connect the node on measuring point, slide rail sleeve and support by BEAM unit;Between support, set up hinge that can only rotate around Z axis;For slide rail sleeve, set up self-contact to realize telescopic movement;Establish discrete unit from measuring point to different support center, for output one-dimensional or two-dimensional compression displacement data and rotation angle data;Calculate two-dimensional displacement component of measuring point relative to reference point.The technical scheme can accurately restore the actual working state of sensor, while ensuring the accuracy of mass distribution during dynamic compression.

This invention relates to the field of radar speed measurement technology, specifically to a method, device, and system for measuring vehicle speed on highways. By acquiring continuous frames of vehicle point cloud data at highway speed measurement points, the three-dimensional shape and spatial position changes of vehicles can be captured. Cluster analysis yields initial clusters, allowing for preliminary classification of different vehicle point clouds. However, vehicle lane changes can cause point clouds to geometrically stick together. Therefore, the motion displacement of vehicle point clouds in adjacent frames is analyzed to determine displacement vectors. Further analysis of their discrepancy characteristics and positional distribution in each frame distinguishes between vehicle clusters and aliased clusters, avoiding misjudgment. Next, within each frame of aliased clusters, clustering based on displacement vectors is used to finely segment the aliased point cloud, restoring the data for each vehicle and eliminating speed measurement interference. Finally, in different frames, the same vehicle is identified by matching the motion and shape characteristics of vehicle clusters. Speed ​​is calculated by matching cluster displacements and frame differences, providing reliable data for traffic management and safety assurance.

The invention belongs to the technical field of aviation safety, relates to an accident reproduction method and system based on multi-fault coupling and QAR reverse driving, and aims to solve the problems of low accident reproduction precision and poor authenticity in the prior art. The method comprises the steps of preprocessing accident flight data to generate an initial fault feature set, and calculating an initial total coupling degree in combination with an aviation fault library; determining a scene weight correction factor based on the initial total coupling degree, performing refinement processing on the flight data, generating QAR time sequence data and a standardized fault feature set, and constructing a multi-fault coupling relation model; according to the QAR time sequence data, the simulator is reversely driven, and a fault injection trigger node is positioned; and when the simulator reproduces to a trigger node, dynamically injecting each fault and the corresponding coupling relationship into the simulator in sequence for dynamic simulation. According to the method, the QAR data refinement processing front closed loop and the multi-fault coupling three-dimensional hierarchical model are constructed, so that dynamic fault injection is realized, and the accuracy and authenticity of data processing and accident reproduction are improved.

The invention relates to the technical field of semiconductor packaging detection, in particular to a semiconductor lead bonding high-precision 3D detection method based on a monocular visible light camera, which comprises the following steps: acquiring an image sequence based on an ultra-low depth-of-field optical system and Z-axis continuous scanning, analyzing by adopting a focusing method after image alignment, and determining an optimal focus point on the surface of a lead layer by layer, a complete three-dimensional model is reconstructed by combining lead trajectory tracking and cubic spline interpolation, and is compared with CAD data to realize high-precision detection of defects such as height, radian, coplanarity, fracture and the like. According to the technical scheme, micron-sized three-dimensional shape reconstruction can be achieved on the premise that only one visible light camera is used, and the device has the advantages of being simplified in structure, high in anti-interference capacity, free of detection blind areas and suitable for on-line detection of vehicle-mounted power devices.

This invention discloses a simulation testing method and system for high-risk scenarios of autonomous driving based on digital twins. The method includes: constructing a three-dimensional continuous traffic environment in a digital twin space based on a real road network topology and multi-source accident data; constructing an accident risk representation object based on the completeness of the multi-source accident data; generating high-risk test scenarios based on the types of accident risk representation objects; converting the high-risk test scenarios into test data corresponding to each functional module in the autonomous driving system, and injecting them into the corresponding functional modules; and outputting high-risk scenario test evaluation results based on the response results of the autonomous driving system. Because this invention constructs a continuous simulation environment by fusing real road network and accident data, and forms accident trigger chains or accident scenario fragments for complete and incomplete accident data respectively, it can comprehensively and accurately recreate various high-risk scenarios, effectively improving the coverage and evaluation accuracy of simulation tests of autonomous driving systems in complex high-risk scenarios.

The application discloses a magnesium alloy micro-arc oxidation film thermal coupling simulation method and system based on Marc, belongs to the technical field of magnesium alloy thermal coupling simulation, and has the advantages that by constructing a magnesium alloy micro-arc oxidation film finite element model, adapting to the form of a Marc solving framework, and then introducing a pulsed ion beam and customized stress response rules, not only the metallurgical bonding state of the raw material base surface is accurately restored, but also the material nonlinear mechanical behavior under high strain rate and wide temperature range in a real environment is compatible, the mechanical response process of the film layer under strong transient thermal shock working conditions is truly reproduced, a transient heat source function of the energy distribution of a quantitative space is constructed, and the dynamic solving of the coupling thermal stress field is realized, the linkage evolution of the temperature field, the stress field and the material phase change ablation behavior is realized, the reliability of the numerical simulation result is improved, and quantitative numerical support is provided for magnesium alloy micro-arc oxidation film ion beam modification process optimization.

The application provides a three-dimensional Gaussian sputtering model training method and device and a dynamic object three-dimensional reconstruction method, and relates to the technical field of three-dimensional reconstruction. The method comprises the following steps: obtaining a training data set, performing Gaussian modeling on the three-dimensional Gaussian data set of a sample dynamic object by using a Gaussian module in an initial three-dimensional Gaussian sputtering model according to color images of multiple perspectives, spattering each three-dimensional Gaussian data in the three-dimensional Gaussian data set to a preset two-dimensional space to obtain corresponding two-dimensional sputtering data, rendering the two-dimensional sputtering data to obtain prediction rendering results of multiple perspectives, calculating a rendering loss according to the color images of multiple perspectives and the prediction rendering images of the corresponding perspectives, obtaining a motion loss according to the color images of multiple perspectives, the three-dimensional Gaussian data set and the sample motion event, and training the initial three-dimensional Gaussian sputtering model to obtain a target three-dimensional Gaussian sputtering model. The application improves the stability and accuracy of three-dimensional reconstruction.