52 results about "Least squares optimization" patented technology

The invention discloses an assembly type composite floor slab cast-in-place section span beam joint design method based on digital twinning, and relates to the technical field of intelligent construction, the method comprises the following steps: adopting a least square optimization algorithm to obtain a material performance correction parameter and a geometric compensation parameter; based on the historical material performance correction parameters and the geometric compensation parameters, training the structure model through a genetic algorithm to generate a digital twin reference model; inputting the material performance correction parameters and the geometric compensation parameters into the digital twin reference model, performing multi-scale finite element analysis, and outputting a multi-scale node performance evaluation report; identifying a multi-scale node performance influence factor and a sensitive area by adopting a response surface method, outputting a design parameter sensitivity analysis result, and converting the multi-scale node performance influence factor into a preliminary node design scheme through a design parameter conversion algorithm; according to the method, the real-time accuracy of node design is remarkably improved through least square dynamic parameter correction.

The invention provides a wind tunnel structure sensitive reflection settlement position intelligent identification method, and belongs to the technical field of wind tunnels. Vibration sensors and displacement sensors are arranged at key positions of a wind tunnel structure to form a monitoring network, collected signals are preprocessed, and a settlement factor matrix is established; a dynamic load matrix is constructed to describe composite load distribution, a vibration burr identification matrix is established, real signals and noise are separated by adopting wavelet transformation, a slow settlement trend matrix is constructed to extract a long-term change rule, and a least square optimization algorithm is adopted to jointly solve each matrix parameter to establish a settlement position identification function. The settlement three-dimensional position coordinate is determined according to the multi-sensor data fusion result, the structure safety state is evaluated through the settlement risk coefficient, and the technical problems that the wind tunnel structure settlement position recognition precision is insufficient, and a real settlement signal and a noise interference signal cannot be effectively distinguished are solved.

The invention relates to a red light therapeutic instrument real-time calibration method based on multi-modal data fusion, which comprises the following steps: firstly, collecting data such as optical power, target surface temperature, internal temperature, reflection spectrum and light source-target surface distance in a time window, and unifying time base alignment; performing anomaly elimination, de-noising and normalization on each modal data, and constructing a standardized data matrix; extracting multi-modal features through power spectral density, wavelet packet decomposition, exponential moving average and Kalman filtering, inputting the multi-modal features into a fusion network containing a modal special encoder and a cross-modal attention unit, and generating a red light treatment state vector; the driving current, the pulse width, the duty ratio, the emission angle and the aging compensation factor are solved in real time through constrained least square optimization and are issued to the driving control module, and the red light source is smoothly adjusted in a soft start mode; the system monitors the output power, calibration is completed if the output power meets a threshold value, and otherwise, the next iteration is started. According to the method, power closed-loop correction can be realized, and dose consistency and thermal safety are improved.

The invention relates to the technical field of point cloud data processing, and discloses a contour feature fine extraction method for workpiece scanning point cloud, which mainly comprises two parts of preliminary extraction and fine processing. In the preliminary extraction part, the point cloud curvature is used as an extraction index, and the calculation method is optimized. Introducing a moving least square optimization normal vector for calculation; normal information is globally considered, normal curvature calculation is optimized in the mode that a normal section circle is constructed through a normal vector, and then the calculation precision of the point cloud principal curvature is optimized. According to the fine extraction part, a spatial clustering algorithm is used for initial feature points, blocky features and linear features are separated while miscellaneous points are removed, and targeted classification processing is carried out according to feature states. For the block-shaped features, using a boundary extraction mode to extract boundary feature points; for linear features, a point cloud thinning mode is used to extract contours. According to the method, the calculation accuracy and stability of the point cloud principal curvature can be optimized, and the extraction precision of the workpiece contour features is improved.

The invention relates to the technical field of wireless communication perception, in particular to a user perception method based on a multi-layer intelligent omnidirectional metasurface, and proposes to realize accurate perception of reflection and refraction user orientations by utilizing heterogeneous characteristics of propagation paths on two sides of a multi-layer intelligent omnidirectional metasurface structure. Based on the architecture, a user reflection and refraction type judgment algorithm based on uplink pilot frequency information is provided, and the user reflection and refraction types are determined based on the heterogeneous characteristics of user channels on the two sides of the multi-layer intelligent omni-directional metasurface and based on least square optimization problem solving and target function size comparison. The propagation path angle estimation algorithm specifically comprises the following steps: firstly, based on uplink pilot frequency information, designing a channel estimation algorithm comprising alternating optimization, constant-rank manifold optimization, convex relaxation and other methods to estimate and obtain a channel matrix by utilizing channel constant-rank and angular domain sparse characteristics; and finally, based on a user type determination result, obtaining user angle information based on each channel matrix by using an orthogonal matching pursuit algorithm.

The invention relates to the technical field of coal mining and rock mechanics numerical simulation, and discloses a stress field reconstruction method of a plane strain numerical simulation model for coal mining, and the method comprises the following steps: firstly, obtaining an actually measured vertical stress increment and an actually measured bearing pressure peak value position of a stope face; then, an initial plane strain numerical simulation model is constructed, the space error between the peak position of the simulated bearing pressure and the actual measurement position meets a preset threshold value by adjusting the excavation step pitch, and therefore the numerical simulation calculation state is locked to eliminate space-time dislocation; a simulation strain increment is extracted in a locked state, a theoretical vertical stress difference equation is constructed based on an elastic mechanics plane strain constitutive relation, and a parameter inversion equation set is established in combination with multi-measurement-point data; and solving the corrected elastic modulus and Poisson's ratio by using a least square optimization algorithm. According to the method, through physical state locking and incremental theory inversion, system errors are effectively eliminated, and the reconstruction precision and mechanical preciseness of the mining-induced stress field are improved.

The invention discloses an intermediate-frequency surface shape white light interference splicing measurement method based on polynomial filtering. The method comprises the following steps: carrying out white light interference splicing measurement on an X-ray plane mirror to obtain a group of interference measurement surface shape results with overlapped adjacent sub-aperture space positions; fitting the surface shape result of each sub-aperture by using a Legendre polynomial to obtain a polynomial coefficient under a preset order; fitting according to a polynomial coefficient to generate a fitting surface shape of a corresponding order, and subtracting the fitting surface shapes of all orders from the measurement result to obtain a group of surface shape results subjected to Legendre polynomial filtering; and splicing the filtered shape results according to a least square algorithm to obtain a final measurement result. The method aims at meeting the splicing measurement of sub-nanometer PV precision required by the intermediate-frequency surface shape error detection of the X-ray plane mirror, the least square optimization splicing method is used, the method is not limited by the low-frequency characteristic of the measurement error of the white light interferometer, and the precision and flexibility of the splicing measurement are improved.

The invention belongs to the technical field of remote sensing image processing, provides a remote sensing image blocking effect elimination method fusing adaptive weight and an anchoring mechanism, and aims to eliminate the blocking effect from a global level, take radiation consistency as a constraint condition, introduce the adaptive weight to carry out differential adjustment on constraint intensity of different overlapped regions, and improve the blocking effect elimination efficiency. Meanwhile, an anchoring mechanism is adopted, stabilization constraint is carried out on an overall solution space, finally, a global equation set is constructed, an iterative least square optimization strategy is used for solving, image radiation consistency and visual continuity are integrally ensured, and the quality of remote sensing images is improved.

A correlation system can be implemented on a user device (e.g., smartphone) to perform image processing tasks e.g., point to point correlations for tasks such as alignment, tracking, and reconstruction of model data. The correlation system can implement normalized cross correlation in a least squares optimization scheme without use of approximations. Normalized cross correlation optimized via least squares can further implement global and local cost functions that are invariant to dynamic lighting conditions.

The application provides a kind of multi-modal feature fusion-based harsh environment image enhancement and restoration method, belongs to image enhancement processing field;The method first obtains multi-modal image and calculates each modal degradation degree graph;Based on the degradation degree and gradient amplitude, a fusion weight is constructed, and the local dominant direction and coherence are obtained by structure tensor analysis;According to the coherence, the pixels are partitioned, and the adaptive processing of each modal gradient is carried out by using hard projection, isotropic fusion or soft projection strategy respectively;Based on the fusion weight, the processed gradient is weighted and fused, and the fusion gradient field is obtained combined with the enhancement factor g (x) ;Finally, an energy function containing pixel fidelity term and gradient fidelity term is constructed, and the enhanced fusion image is reconstructed by weighted least squares optimization;The application suppresses the contribution of blurred modal through degradation perception mechanism, and solves the gradient direction conflict through structure adaptive projection, which significantly improves the clarity and structure fidelity of multi-modal image fusion in harsh environment.

This application discloses a deconvolution method for gas chromatography-mass spectrometry (GC-MS) data, relating to the field of GC-MS data analysis technology. This method utilizes a non-negative matrix factorization function to initially decompose the time-mass-charge ratio two-dimensional intensity matrix, obtaining an initial elution curve matrix and an initial mass spectrometry matrix. An exponentially modified Gaussian peak shape modeling function is then used to fit the initial peak shape parameter set. For each component: based on the initial peak shape parameter set, an exponentially modified Gaussian peak shape modeling function is used to generate the original elution curve matrix. Non-negative least squares optimization and non-linear least squares optimization methods are used to iteratively optimize the mass spectrometry matrix, peak shape parameter set, and elution curve matrix. Finally, a comprehensive scoring function is used to select the optimal component set, and the optimal peak shape parameter set, mass spectrometry matrix, and elution curve matrix for the optimal component set are sorted and output. This application enables precise separation of co-eluted components, improving the accuracy and reliability of complex mixture analysis.

The invention discloses a GIS partial discharge rapid positioning method based on a multi-normal projection principle, and relates to the technical field of partial discharge on-line monitoring, a three-dimensional photoelectric array formed by silicon optical semiconductor sensors is used for obtaining a partial discharge signal amplitude, a direction vector is constructed, and a unit direction is obtained through solving a local intersection point. And further combining with least square optimization of multi-unit rays to obtain initial position estimation of the partial discharge source. On the basis, an amplitude correction and iteration updating mechanism is introduced, observation data are corrected, and a real position is gradually approached, so that high-precision three-dimensional partial discharge source positioning in a complex electromagnetic environment is realized. Through the combination of the silicon optical sensor three-dimensional array and the improved multi-vector projection iterative algorithm, high-precision partial discharge positioning can be realized in a complex electromagnetic environment, the operation reliability and the operation and maintenance efficiency of equipment are remarkably improved, and the method has a wide engineering application prospect.

The invention discloses a double-block type sleeper laying fine adjustment system and method based on machine vision, and relates to the field of rail traffic engineering.The method comprises the steps that firstly, embedded metal measuring nails are arranged along a rail to serve as physical reference marks, and a stable three-dimensional coordinate system is constructed through back measurement, least square optimization and dynamic remeasurement; then, machine vision is used for collecting sleeper height, transverse center line position and pitch angle data in real time, and the corresponding relation between adjusting operation and parameters is synchronously recorded; kalman filtering noise reduction is carried out on the collected data, a normalized deviation cumulant is calculated based on a sliding window, and an error accumulation trend is identified through a continuous increasing situation; and finally, extracting high-frequency operation in an error accumulation period, calculating a unit operation quantity deviation change rate, and executing a differential adjustment strategy, including independent parameter adjustment, reverse compensation operation or pause operation, according to the unit operation quantity deviation change rate.

A bone registration method and system for joint replacement surgery, based on a VSF surface in a preoperative coordinate system generated from preoperative bone image data and a source point set SPS in an intraoperative coordinate system generated from point data collected from the patient's actual bone surface during surgery, determines the transformation relationship between the preoperative and intraoperative coordinate systems through initial and precise registration. The method includes: a first step: selecting multiple corresponding point pairs from the source point set SPS and the VSF surface, performing initial registration based on Singular Value Decomposition (SVD), and obtaining the initial transformation matrix; and a second step: performing precise registration based on the initial transformation matrix using the ICP method. Each ICP iteration employs a least-squares optimization method that minimizes the sum of squared distances between each point in the transformed source point set SPS and the corresponding projection points in the projection point set PSPS. The algorithm stops when the error is less than a set value or the number of iterations reaches a set value.

This invention discloses a high-precision density reversal method and system for gravity anomalies based on regular, stable, and high-speed solution. [Solution] Based on the above-mentioned field source regularization solution, we propose using data with false anomalies at the field source imaging depth as boundary value conditions. Again, by solving the upper space of the first class boundary value problem of Laplace's equation, we calculate gravity anomalies on the ground and perform least-squares optimization iterative solution with the actual ground anomalies to obtain high-precision anomaly imaging data at the field source depth that does not contain false anomalies due to the Gibbs effect. This depth is used as the top interface of the field source, and the top interface is divided into grid prisms to construct the field source morphology. Density inversion is performed on each prism top surface in the frequency domain to obtain the density distribution of the field source morphology. Since this algorithm is implemented in the frequency domain and the inversion is performed at a known field source depth, the present invention is a gravity density inversion method with high speed, high precision, and high resolution characteristics, and provides a new concept for realizing the solution of gravity field source density parameters.

This invention provides a method and apparatus for measuring the articulation angle of a semi-trailer based on multi-sensor fusion, comprising: taking the absolute value of the difference between a second distance and a third distance to obtain the compensated distance difference between two ultrasonic sensors; determining the distance residual based on the compensated distance difference between the two ultrasonic sensors and the articulation angle formula; adding the distance between the projection of a 3D feature point in the world coordinate system onto the plane of the left image and its corresponding feature point in the left image, and the distance between the projection of the feature point onto the plane of the right image and its corresponding feature point in the right image, to obtain the reprojection residual; determining the overall residual based on the distance residual and the reprojection residual; constructing a least-squares optimization objective function based on the overall residual, and solving iteratively using the Levenberg-Marquardt algorithm. This invention provides a method and apparatus for measuring the articulation angle of a semi-trailer based on multi-sensor fusion to solve the problems of environmental sensitivity, poor stability, and difficult installation of contact sensors in single-sensor solutions.

The application discloses a kind of residual stress detection method, system and storage medium based on optical fiber distributed sensing blind hole method, belong to residual stress test and optical fiber sensing technical field.The method first establishes the strain release coefficient database under different hole depth and radial position by finite element analysis, and constructs the continuous correction factor function model about dimensionless hole depth and radial position for correcting the difference between blind hole and through hole.Implementation, distributed fiber optic sensor is laid on the surface of workpiece and blind hole is drilled, and the release strain data of multiple angles and multiple radial positions before and after drilling are collected.The measured data is substituted into the theoretical strain model corrected by the correction factor, and the residual stress parameters are obtained by least square optimization inversion.The application does not need to rely on sample calibration, and the measurement accuracy and applicability of blind hole method in actual engineering are significantly improved by accurate matching of full-field strain data and theoretical model.

A method and system for generating and displaying a low frequency model of a seismic survey area is provided. The method and system include defining a seismic survey geometry of a seismic survey; processing the seismic data to generate superimposed seismic data and log data to obtain elastic properties; importing the superimposed seismic data and the processed log data into a defined seismic survey geometry; generating envelope data using the superimposed seismic data; generating a low-frequency seismic trace for each well; calculating a least square optimization coefficient model of each well position according to the generated envelope data and the low-frequency seismic trace of each well; interpolating a coefficient model into the seismic survey geometry using a covariance technique and the imported superimposed seismic data; and generating a three-dimensional low frequency model for display by inversion using the envelope data and the interpolated coefficient model.

The application discloses a kind of based on polynomial filtering's intermediate frequency surface shape white light interference splicing measurement method, including to X ray plane mirror is carried out white light interference splicing measurement and obtains a group of adjacent sub-aperture spatial position overlapping interference measurement surface shape result;Using Legendre polynomial, the surface shape result of each sub-aperture is fitted to obtain polynomial coefficient under preset order;According to polynomial coefficient fitting, the fitting surface shape of corresponding order is generated, and all order fitting surface shapes are removed from measurement result together, obtain a group of after Legendre polynomial filtering surface shape result;After filtering surface shape result is spliced according to least square algorithm to obtain final measurement result.The present application aims to meet the sub-nanometer PV precision splicing measurement required in X ray plane mirror intermediate frequency surface shape error detection, using least square optimization splicing method and not being restricted by white light interferometer measurement error low frequency characteristics, improve the precision and flexibility of splicing measurement.

The application provides a metal component thickness measurement method based on eddy current detection, comprising: building an eddy current sensor with coaxial excitation and two receiving coils, forming two pairs of excitation-receiving coil units with different lift-off distances, and measuring the eddy current response signal of the metal component through a sweep frequency eddy current mode; calculating the mean value of the response signal, and obtaining a differential mutual inductance response curve by subtracting the mean value; analyzing the frequency and mutual inductance value corresponding to the intersection point of the curve; establishing an eddy current response analytical theory model, setting the electromagnetic parameters of the metal component and the geometric parameters of the sensor; inputting the frequency and mutual inductance value of the intersection point into the analytical model, converting the thickness measurement problem of the metal component into a least square optimization problem, using an improved Newton iteration algorithm to invert the thickness until the convergence condition is met, and finally outputting the thickness measurement result of the metal component. The method can realize accurate measurement of the thickness of the metal component and is not affected by the fluctuation change of the lift-off distance.

This invention relates to the field of power grid sensing and data analysis technology, and discloses a two-stage topology identification method, device, equipment, and storage medium for power distribution systems, which improves the accuracy and reliability of power distribution network topology identification. The two-stage topology identification method for power distribution systems includes: acquiring smart meter measurement data and cumulative energy consumption data of inter-meters, and performing preprocessing through current calculation, time alignment, and power conversion; subsequently, constructing a constrained least squares optimization problem to solve for the impedance sub-matrices between smart meters and between smart meters and inter-meters; based on the impedance sub-matrices, using a recursive grouping algorithm to construct a smart meter subgraph, and then successively determining the connection positions of inter-meters to generate the target power distribution network topology.

The invention discloses a multi-modal image positioning and calibration analysis method and system based on a magnetic induction array, relates to the technical field of data analysis, and provides a magnetic induction array formed by magnetic sensors. Preoperative images are collected, a three-dimensional model of an operative area is constructed based on the images, a stimulation target area and a navigation path are planned, and a preoperative planning point set is generated; a datum point is selected, rigid registration with an actual datum point in the operation is completed, and an initial coordinate mapping relation is established; based on the boundary of a contact operation target area, a probe magnetic excitation coil excites a magnetic field, an array sensor collects three-axis magnetic field data according to a preset sampling period, and an intraoperative actual measurement point set is generated; inputting an actual measurement point set acquired by a sensor into an LM solver, and outputting probe coordinates in real time; and performing coordinate correction based on a deformation compensation mechanism, and performing real-time calibration by the system according to the corrected coordinates. According to the method, the non-rigid deformation deviation of the tissue in the operation can be effectively corrected by combining an affine transformation matrix and a least square optimized deformation compensation mechanism.

The present invention discloses a method and system for high-precision density inversion of gravity anomalies based on regular stable fast solution. [Solution] Based on the above source regularization solution, we propose using data containing spurious anomalies at the source imaging depth as boundary value conditions. Then, through the upper half-space solution of the first-class boundary value problem of Laplace's equation, we calculate the gravity anomaly on the ground, and perform a least-squares optimization iteration with the actual ground anomaly to obtain high-precision anomaly imaging data at the source depth without spurious anomalies caused by the Gibbs effect. This depth is used as the top interface of the source, and the top interface is divided into grid prisms to construct a source configuration. Density inversion is performed on each prism top surface in the frequency domain to obtain the density distribution of the source configuration. Because this algorithm is implemented in the frequency domain and the inversion is performed at a known source depth, the present invention provides a gravity density inversion method with high speed, high precision, and high resolution, and provides a new concept for solving gravity source density parameters.

The application discloses a scene-based multi-scale non-uniformity correction method, which firstly establishes an image pyramid for an input image; for each layer of the image pyramid, a local linear model in guided filtering is used, i.e., an image after non-uniformity removal is taken as a guide image, in a local neighborhood window, the guide image and a filtering output image are in linear relationship, meanwhile, a low-frequency non-uniformity model in the local neighborhood window is considered as a polynomial model with coordinate axes as variables; then a weighted least square optimization method is used to constrain the non-uniformity, the original optimization problem is decomposed into two sub-problems, the alternating direction multiplier method is used to solve the two sub-problems, iteration optimization is carried out, and the non-uniformity of the image is obtained; finally, the non-uniformity obtained from each layer of the image pyramid is reconstructed, a final non-uniformity image is obtained, and a clear image is obtained by subtracting the non-uniformity from the original image.

The application provides a kind of relative attitude autonomous measurement method between multiple space vehicles based on space-based radar precision measurement.Space formation space vehicles are all equipped with space-based radar, and can measure relative distance and angle, which is converted into LOS observation vector in each space vehicle body coordinate system;Since the same physical line-of-sight vector in different space vehicle body coordinate systems satisfies the rotation matrix transformation relationship, multiple non-collinear LOS vectors are obtained through mutual observation between multiple space vehicles, and a least squares optimization problem is constructed;The optimal rotation matrix is solved by singular value decomposition, and the result is constrained to SO (3) group to ensure the legality of the rotation matrix, and the complete three-axis relative attitude between space vehicles is finally solved.The application does not need to know the relative attitude and position of the space vehicle as priori, and can realize the relative attitude measurement between any two space vehicles through distance and angle measurement values.

The invention discloses a bamboo strip character denoising method based on a fractional order diffusion equation inverse problem POD algorithm, and belongs to the field of image processing and inverse problem calculation, and the method comprises the steps: carrying out the preprocessing of a bamboo strip image, and obtaining a terminal observation image; based on the set diffusion time and the fractional order, a finite difference method is adopted to solve a positive problem of a time fractional order diffusion equation, and a snapshot data set is generated; based on the snapshot data set, a singular value decomposition method is adopted to extract a main mode, and a reduced-order model is established; based on the terminal observation image and the reduced-order model, constructing a regularization least square optimization problem and adopting a gradient iteration algorithm for solving, and performing inversion to obtain a clear image at an initial moment; and processing the clear image at the initial moment by adopting a threshold segmentation method, and outputting a finally recovered high-definition bamboo strip image. The invention provides an efficient and stable treatment means with clear physical significance for digital protection of the fragile cultural relics such as the bamboo strips.

The invention relates to a multi-radar registration and association method based on minimum residual error, which comprises the following steps: acquiring position and speed measurement data of M radars to N targets, traversing all possible association conditions, constructing a position and speed residual error function for each condition, and performing weighted summation to form a total target function; and solving a minimum residual value by using a least square optimization algorithm, finally selecting an association result corresponding to a global minimum residual as an optimal solution, and synchronously outputting registration parameters between radars and a real motion state of a target. According to the method, the joint optimization solution of the registration and correlation coupling problem under the condition of no prior information is realized.

The invention belongs to but not limited to the technical field of satellite communication, and particularly relates to an OTFS-based fractional Doppler channel estimation method and system in low-orbit satellite communication, and the method comprises the steps: S1, employing a ZC sequence as a pilot frequency, and providing a low-PAPR DD domain pilot frequency structure; s2, analyzing the influence of IDI on a ZC pilot frequency period correlation result, and revealing a module value distribution rule; s3, initial estimation based on correlation peak value calculation is provided, and initial channel parameters can be obtained; and S4, constructing a non-convex constraint least square optimization problem, and performing solution by using an SQP algorithm so as to obtain more accurate channel parameters. According to the DD domain frame structure provided by the invention, the ZC sequence is adopted as the pilot frequency, and the guard interval is set, so that the PAPR can be reduced, and the interference between data and the pilot frequency can be effectively reduced. The module value distribution rule of the correlation result of the pilot frequency period of the ZC sequence under the influence of IDI is disclosed for the first time, and the rule is fully utilized to carry out initial channel estimation, so that the accuracy and reliability of estimation are improved.

The invention discloses a multi-machine cooperative progressive interference source searching method based on received signal strength indication. The method comprises the following steps: inputting configuration information such as a searching area and unmanned aerial vehicle parameters; planning the flight direction of the unmanned aerial vehicle according to the previous multi-vehicle fusion interference source position, and collecting RSSI data during flight; taking a minimum sampling and prediction RSSI difference value as a target, and obtaining an initial interference source position of a single machine through least square optimization; calculating a measurement confidence coefficient by combining prediction of an RSSI difference value and a measured value, and performing time sequence fusion on single-machine multi-section position estimation results; and carrying out multi-unmanned aerial vehicle space fusion on the time sequence fusion results of all the unmanned aerial vehicles to obtain the interference source position estimation of the section. According to the invention, multiple unmanned aerial vehicles are utilized to cooperatively carry out interference signal intensity acquisition, and high-reliability and high-adaptability search of an interference source is realized in a staged and successive approximation mode.

This invention discloses a dexterous hand telemanipulation method based on grasping motion priors, comprising: training GraspNet to generate target grasping posture parameters of the human hand; generating a ground truth trajectory based on the initial posture and target posture through least squares optimization, and constructing a noisy reference trajectory; redirecting the human hand's MANO parameters to a five-fingered manipulator; collecting a multimodal grasping trajectory dataset containing visual, proprioceptive, reference posture sequences, and ground truth posture sequences in the Isaac Gym simulation environment; extracting multimodal features and using a multi-layer Transformer network to fusion and predict the ground truth manipulator posture at the current moment; and training a motion prior network using mean squared error loss. This invention introduces grasping motion priors into telemanipulation, which can effectively correct trajectory deviations caused by perceptual noise, significantly improve the grasping success rate, and enhance the generalization ability from simulation to real-world scenarios.