85 results about "Vector decomposition" patented technology

The invention discloses a wind field data modeling method and system for a small and medium-sized unmanned aerial vehicle wind resistance test, and the method comprises the steps: collecting the topographic data, wind speed data and fan data of a test wind field, and determining a preset region and an unmeasured region according to the preset trajectory of an unmanned aerial vehicle based on the test wind field; the method comprises the following steps: acquiring boundary conditions of an unmeasured area and disturbance grid wake flow through fluid simulation based on topographic data and a preset area, performing wind speed vector decomposition on a grid according to a preset wind field boundary condition, and acquiring a preset wind field model according to a disturbance time sequence, an unmanned aerial vehicle maneuvering weight and the disturbance grid wake flow, and obtaining a predicted wind field of the unmeasured area through Gaussian process regression according to the boundary condition of the unmeasured area based on a preset wind field model, and coupling the predicted wind field through a grid time sequence by using different unmanned aerial vehicle preset trajectories to obtain a target wind field model. According to the method, through space-time collaborative interpolation and fluid simulation of Gaussian process regression, the inlet boundary condition better conforms to the fluid mechanics law, and the physical consistency and reliability of the target wind field are improved.

The invention provides a dialogue processing method and system based on a large model, and the method comprises the steps: obtaining a dialogue text length value, employing a pre-established semantic boundary identification set to detect a topic turning frequency value according to the dialogue text length value, and combining a historical statement sequence sorted according to time in a context window, generating an initial semantic segmentation vector containing feature extraction dimensions; according to the initial semantic segmentation vector, calculating a topic coherence score by adopting a dependency weight distribution table, and separating sub-vectors of which the orthogonality degree is higher than a preset threshold value through a vector decomposition precision value to generate a preliminary structured representation matrix; extracting emotional intensity fluctuation features and knowledge density distribution features from the preliminary structured representation matrix, and generating a refined structured representation matrix after adjusting the rank number of the matrix; and calculating the Euclidean distance between the topic coherence sub-vector and the logical reasoning sub-vector in the refined structured representation matrix.

The invention relates to the technical field of automatic laying of composite materials, in particular to an automatic compensation method, system and equipment for errors between silk laying belts and a storage medium. The method for compensating the error between the automatic silk laying belts comprises the following steps: establishing a mathematical mapping relation between the error compensation amount between the belts and robot control parameters through a standard sample piece laying test, and forming an error compensation theoretical model; the method comprises the following steps: aiming at a robot type fiber placement machine structure, constructing an inter-band error model based on pose kinematics modeling, robot and additional axis multi-axis interpolation modeling and error vector decomposition; the tail end target position, the attitude angle and the error compensation amount in the laying program are read in real time, and real-time control input is obtained; and on the basis of the error model, the corrected tail end position is inversely solved into a real-time adjustment angle of each joint, a robot inverse solution algorithm is corrected, and a joint control instruction is generated. According to the automatic tow belt laying device, the laying position precision of the automatic tow belt laying can be remarkably improved.

The invention discloses a hand-eye calibration method suitable for a five-axis series-parallel mechanical arm, which comprises the following steps of: analyzing coordinate conversion error distribution characteristics according to an initial hand-eye calibration matrix calculation result, and separating a total error into two components, namely a translation error and a rotation error through an error vector decomposition algorithm, obtaining distribution rules and variation trends of the errors in different spatial directions; establishing a calibration precision evaluation system according to the verification result, calculating the repeated positioning precision and the absolute positioning precision of the system by counting and analyzing the position deviation data of multiple tests, and obtaining the system performance index in the current calibration state; a self-adaptive updating mechanism is established for system performance indexes, local updating of calibration parameters is automatically triggered when environment temperature change or small deformation of a mechanical structure is detected, and a compensation model is adjusted in an incremental learning mode to avoid comprehensive recalibration.

The invention discloses a face gear worm grinding machine error compensation method based on sensitivity analysis and a multi-source error mapping model, which comprises the following steps of: 1, constructing a multi-source error-tooth surface error mapping model based on a vector decomposition and truncation function method, the truncation function method is used for quantitatively describing propagation mechanisms of geometric errors and thermal errors in a machining chain and influences of the propagation mechanisms on tooth surface deviation, and high-order error terms are eliminated to improve calculation efficiency; 2, responding to the output of the mapping model, applying an improved Morris sensitivity analysis method, distributing a weighting coefficient higher than a geometric error component for a thermal error component, and calculating a weighting sensitivity coefficient so as to identify a key error component which has the greatest influence on the tooth surface precision; and 3, based on the key error component, a motion axis compensation value is generated through an error compensation model, and the compensation value is input into a machine tool control system to adjust the relative position of the worm grinding wheel and the workpiece, so that the tooth surface error is reduced.

The invention relates to the technical field of geological engineering safety risk evaluation, in particular to a regional tectonic stress risk quantitative analysis method for tunnel engineering. The method comprises the following steps: collecting multi-source geomechanical parameters, simulating and inverting a tectonic stress field, and verifying the tectonic stress field; constructing a multi-scale geomechanical model, coupling simulation tectonic stress and gravity stress, and generating a stress space basic layer; utilizing Kriging interpolation and inversion decomposition to obtain a stress value and a directional diagram layer; calculating an included angle between the maximum principal stress direction and the tunnel axis, and performing vector decomposition to obtain a vertical stress component; and carrying out standardized normalization on the factors, determining a weight calculation risk value, and finally generating a tectonic stress risk quantitative grading graph and carrying out GIS visual rendering. The tunnel engineering-oriented risk quantitative analysis mechanism is constructed by retaining the directivity and spatial heterogeneity characteristics of the tectonic stress tensor, and the authenticity, resolution and engineering applicability of tectonic stress identification are remarkably improved.

The invention provides a three-dimensional flow velocity field inversion method and system based on coherence-assisted SAR adaptive offset tracking, and a storage medium, so as to improve the problem of decoherence caused by time heterogeneity in a synthetic aperture radar interferometry (InSAR) technology, and relates to the technical field of electric power engineering geological disaster monitoring. The method comprises the following steps: establishing a space-time baseline network between SAR images; obtaining coherence data after geocoding; selecting a pair of SAR images from the space-time baseline network as a reference unit, calculating an offset result through a pixel offset tracking technology for the reference unit, and determining an optimal matching window parameter of the reference unit; establishing a mapping function between coherence and optimal matching window parameters; according to the established mapping function, calculating an optimal matching window parameter of each remaining SAR image pair; and in combination with the line-of-sight average deformation rate fields of the ascending and descending rails, inverting a surface three-dimensional flow velocity deformation field of the target area through a geometric vector decomposition model.

The invention provides an automatic calling method for a welding parameter subprogram based on a three-dimensional model. The automatic calling method comprises the following steps: (1) establishing a welding process parameter library comprising a welding process parameter table and a process parameter matching rule table; (2) on the basis of the established three-dimensional model, identifying structural parameters among the parts, and generating corresponding welding seams; (3) in the established three-dimensional model, performing three-dimensional marking on the generated welding seam to obtain a welding seam section code; the information of the three-dimensional label is process parameter information including the size of a weld fillet, the arrangement of weld layer channels and the welding position; (4) taking the obtained weld joint section code as an initial condition value, obtaining corresponding welding process parameters from a welding process parameter library, and realizing automatic calling of the welding parameters; and aiming at the obtained welding process parameters, under the world coordinate system of the welding robot, according to the vector decomposition of the Y axis in the defined offset coordinate system in the Z direction under the world coordinate system, the A direction and the B direction of the welding parameter subprogram are automatically judged.

The invention belongs to the technical field of electric power measurement, and discloses a calibration method and system of an intelligent measurement switch and a multi-switch synchronous measurement method, and the method comprises the steps: obtaining voltage and current waveform data of the intelligent measurement switch under a nonlinear load condition, and extracting a frequency spectrum feature matrix of the voltage and current waveform data; constructing an amplitude-phase coupling error model representing a nonlinear incidence relation between amplitude fluctuation and phase deviation, inputting the spectrum characteristic matrix into the amplitude-phase coupling error model, and calculating an amplitude-phase coupling error vector; performing vector decomposition on the amplitude-phase coupling error vector, and separating out an in-phase component error and an orthogonal component error; and generating a calibration compensation coefficient according to the in-phase component error and the orthogonal component error, and performing vector calibration on the intelligent measurement switch, thereby improving the calibration precision of the intelligent measurement switch.

The invention relates to the technical field of pipeline corrosion detection, in particular to a method for separating corrosion defect magnetic anomaly signals of a pipeline with a coating layer. The method comprises the following steps: acquiring a magnetic induction intensity frame sequence of the surface of a pipeline with a coating layer, performing normalization processing on data, extracting multi-scale features through wavelet decomposition, and calculating a signal-to-noise ratio, a space variance and an abnormal area proportion of an abnormal magnetic induction area; then, a target magnetic induction intensity frame combination with the highest information entropy is screened out from a large amount of data through a genetic algorithm; performing magnetic vector decomposition, wavelet packet transformation and singular value decomposition on the target frame, and constructing a continuous magnetic anomaly distribution map; and finally, calculating the three-dimensional parameters of the corrosion defect through a pre-calibrated sensor response model and an inversion algorithm. According to the method, the data processing amount is remarkably reduced while the detection precision is ensured.

The invention discloses an oval scanning laser radar wave refraction correction error correction method, which belongs to the technical field of airborne laser sounding, is used for airborne laser radar underwater topographic survey wave refraction correction, and comprises the following steps of: performing filtering processing on sea surface point cloud by adopting a statistical outlier removal filtering method; and extracting a sea surface gradient normal vector corresponding to each point by adopting an optimized moving least square method, simplifying the measurement process of the oval scanning laser radar to be carried out under a carrier coordinate system, constructing a refraction correction model by utilizing a Snell law and a vector decomposition method, and obtaining a corrected seabed point cloud coordinate. According to the method, the sea surface point cloud data is denoised, the actually measured sea surface point cloud is fitted through the moving least square method, polynomial fitting is replaced with covariance matrix characteristic decomposition, calculation is easy and convenient, polynomial fitting errors are avoided, the process is greatly simplified through vector decomposition calculation, and only a single solution exists.

The invention discloses a symbol coding and vector decomposition method based on a super-dimensional calculation model, and belongs to the technical field of neural symbol models and super-dimensional calculation. According to the method, a symbol coding method is used, and an embedding space of additional information is added in coding, so that representation of multiple subclass levels is supported; an efficient decomposition method is designed, redundant classes are selectively eliminated by identifying labels of target classes, the anti-noise capability and mathematical characteristics of super-dimensional vectors are fully utilized, the calculation efficiency and accuracy when multiple objects with class-subclass relations are expressed and decomposed are remarkably improved, and the method is suitable for large-scale popularization and application. And the limitation of an existing super-dimensional calculation model is overcome.

The invention discloses a digital self-interference cancellation method and device based on an enhanced neural network structure, and the method comprises the steps: carrying out the complex value splitting operation and the non-linearity removal operation of a first stepped grid structure, obtaining a second stepped grid structure, carrying out the complex value splitting operation and the non-linearity removal operation of a first sliding window grid structure, and obtaining a second sliding window grid structure; obtaining a second sliding window grid structure; obtaining a first output of the dynamic phase layer according to the second stepped grid structure or the second sliding window grid structure; constructing a vector decomposition structure to separate amplitude information and phase information; feeding the amplitude information into the nonlinear hidden layer to obtain a second output; fusing the phase information and the second output; feeding the fused data into a full-connection output layer, and outputting a simulation result; and according to the simulation result, performing self-interference cancellation operation on the received signal to obtain a target signal. The method can improve the digital self-interference cancellation performance and reduce the network complexity, and can be widely applied to the technical field of full-duplex communication.

The invention discloses a plateau frozen soil thickness detecting and monitoring method and system based on distributed optical fiber acoustic sensing. The method comprises the steps that a spiral optical fiber and a linear optical fiber are arranged in frozen soil according to a preset observation position, and vibration signals are collected through the spiral optical fiber and the linear optical fiber respectively; the collected vibration signals are preprocessed, and the vibration signals are converted into particle velocity signals; vector decomposition is carried out on two mass point velocity signals obtained by the spiral optical fiber and the linear optical fiber, and a surface wave component is inhibited to enhance a mass wave component; carrying out self-excitation and self-receiving recording on the measuring point position according to the data after vector decomposition; recording data and picking up frozen soil upper and lower limit travel time; according to the frozen soil upper and lower limit travel time, the overlying stratum speed and the frozen soil layer speed, obtaining the frozen soil upper limit burial depth and the frozen soil thickness. The device can output the spatial distribution condition of the frozen soil in real time, monitors the buried depth, thickness and upper and lower limit positions of the frozen soil in real time, improves the geological interpretation effect for railway investigation and design, and provides a scientific basis for safe operation of the railway.

The invention discloses a lightweight bridge and dynamic task vector federated multi-task learning method, which comprises a federated learning system, the system comprises a plurality of clients and a central server, and the method comprises the following steps: 1) initialization and distribution; 2) performing local training and processing on the client; 3) performing dynamic task vector decomposition and compression transmission; 4) aggregation and reconstruction of the central server side; and 5) updating and issuing the model. According to the method, the communication efficiency is improved in order of magnitude, the communication bottleneck is relieved fundamentally, the personalized performance of the model is remarkably enhanced and exceeds that of an existing baseline method, inter-task interference is effectively restrained, training stability and convergence are improved, excellent balance between communication overhead and calculation / personalized performance is achieved, and good adaptability and robustness are achieved.

The invention provides a second-hand vehicle data set construction method and device based on data governance and a storage medium, and the method comprises the steps: obtaining an original vehicle description text in a second-hand vehicle transaction scene, carrying out the text cleaning and standardization processing, obtaining a to-be-processed vehicle description text set, carrying out the event unit extraction of the to-be-processed vehicle description text set, and obtaining a to-be-processed vehicle description text set; the method comprises the following steps: identifying and extracting event units containing operation behaviors and state changes in a to-be-processed vehicle description text set, performing double-base vector decomposition processing on each event unit to obtain double-base vectors, embedding the double-base vectors into a preset three-dimensional event space, detecting the double-base vectors in the three-dimensional event space, and obtaining a vehicle description text set; and identifying and correcting the event units with contradictions, generating a correction vector set for eliminating the contradictions, projecting the correction vector set to a preset second-hand vehicle valuation model data field space, and generating a structured data unit aligned with the second-hand vehicle valuation model data field space.

The method for quickly inverting aircraft wake vortex parameters based on mobile platform laser remote sensing comprises the following steps: a laser radar system scans a wake vortex influence area, and measurement data are subjected to quality control processing; a wake vortex core position and a vortex core radius are estimated; a wind speed of an arbitrary measurement point A located in the wake vortex range is subjected to vector decomposition, and the wind speed is expressed as a function of a wake vortex circulation parameter W 、 V 、 U A background wind speed and a laser radar motion speed are eliminated by using a non-wake vortex area measurement point; a difference between radial velocities of two measurement points in and out of a wake vortex influence range in the same laser emission direction is expressed as a function of a wake vortex circulation; and a turbulent disturbance is weakened by multiple measurements in the wake vortex influence range, and then a least square method is used to fit and solve the wake vortex circulation. The present application can accurately invert and correct aircraft wake vortex parameters by using less observation data. The calculation result is fast and accurate, and can provide scientific and reliable data support for aviation safety and aircraft cooperative operation.

The embodiment of the present application provides a bearing stiffness detection method, device and storage medium, the method comprising: obtaining the force exerted on each bearing shell on the bearing during the target motion cycle; determining the vector trajectory of the resultant bearing force during the target motion cycle based on the force, and performing vector decomposition on the vector trajectory to obtain the load at each sampling moment; obtaining the motion trajectory of the axis relative to the origin during the target motion cycle, and performing vector decomposition on the motion trajectory to obtain the motion components at each sampling moment; obtaining the bearing stiffness at each sampling moment during the target motion cycle based on the load and motion components. In the embodiment of the present application, the motion trajectory is obtained by a distance sensor group and the vector trajectory is obtained by a force sensor, and the bearing stiffness is then calculated based on the two trajectories, which can ensure the accuracy of the bearing stiffness detection.

The invention provides a synchronous modulation phase alignment method and device, and belongs to the technical field of motor operation control. The method comprises the steps that when phase deviation is larger than a set deviation threshold value, a second phase alignment mode is started, and the output frequency of actual output voltage is equal to the expected frequency of expected output voltage; determining a plurality of sampling points according to a preset selection principle; recording an angle between two adjacent output vectors in synchronous modulation as shown in the specification, recording an angle between output vectors in an adjustment process as shown in the specification, and recording an angle deviation between an adjustment starting vector and an adjustment ending vector as shown in the specification; determining the number of intermediate vectors according to the adjustment ending vector and the carrier direction of the jump point; obtaining the duration of the vector through vector decomposition; determining a carrier direction according to a carrier diversity principle, and determining a clamping mode according to sector division; and determining a vector arrangement sequence according to the carrier direction and the clamping mode. In this way, no deviation exists between the output vector and the expected vector in the adjusting process, and the adjusting speed is increased.

The invention discloses a moving track prediction method and a related device, which are applied to scenes such as artificial intelligence and games. The method comprises the steps of representing object state information, game environment information and game task information and features associated with a target object in a first target time period in a target game as object state features, game environment features and game task features; performing hybrid coding to obtain a prediction output vector of the target object in a second target time period after the first target time period in the target game; and decomposing each prediction output vector into a prediction mean vector, a prediction standard deviation vector and a prediction probability vector so as to accurately determine a prediction moving track of the target object in the second target time period in the target game. Even if the movement track of the player in the game is very complex, the method can accurately predict the movement track of the player in the game in the future time period, so that the prediction effect of the movement track in the game is improved.

The present invention discloses a method and system for detecting and monitoring the thickness of permafrost on the plateau based on distributed fiber optic acoustic sensing. The method includes deploying spiral and linear optical fibers in the permafrost according to predetermined observation locations, collecting vibration signals using the spiral and linear fibers, respectively; preprocessing the collected vibration signals to convert them into particle velocity signals; vector decomposing the two particle velocity signals obtained from the spiral and linear fibers to suppress the surface wave component and enhance the body wave component; self-excited and self-received recording of the measurement point locations based on the vector decomposition data; extracting the upper and lower limit travel times of the permafrost from the recorded data; and obtaining the upper and lower limit burial depth and thickness of the permafrost based on the upper and lower limit travel times of the permafrost and the velocity of the overlying strata and the permafrost layer. The present application can output the spatial distribution of permafrost in real time, monitor the buried depth, thickness, and upper and lower limit positions of the permafrost in real time, and provide a scientific basis for improving geological interpretation for railway survey and design and ensuring safe railway operation.

The application provides a synchronous modulation phase alignment method and device, and belongs to the technical field of motor operation control. The method comprises the following steps: when the phase deviation is greater than a set deviation threshold, a second phase alignment mode is started, and the output frequency of an actual output voltage is equal to the expected frequency of an expected output voltage; a plurality of sampling points are determined according to a preset selection principle; the angle between two adjacent output vectors of the synchronous modulation is denoted as, the angle between the output vectors in the adjustment process is denoted as, and the angle deviation between the adjustment starting vector and the adjustment ending vector is denoted as; the number of intermediate vectors is determined according to the carrier direction of the adjustment ending vector and the jump point; the duration of the vector is obtained through vector decomposition; the carrier direction is determined according to the carrier difference principle, and the embedding mode is determined according to the sector division; and the arrangement sequence of the vector is determined according to the carrier direction and the embedding mode. In this way, there is no deviation between the adjustment process output vector and the expected vector, and the adjustment speed is accelerated.

The invention provides a federated learning robust aggregation method and system for resisting backdoor attacks, and belongs to the technical field of network and information security. The method comprises the following steps: acquiring model update vectors submitted by a plurality of clients, and decomposing the model update vectors into a plurality of hierarchical update data according to the hierarchical relationship of a global model; according to each hierarchy update data of each client, extracting a multi-dimensional feature of the client; determining a global total abnormal score of the client according to the multi-dimensional features of the client; and calculating a comprehensive reputation of the client according to the global total abnormal score, determining an aggregation weight according to the comprehensive reputation, aggregating updating data of each level of the client according to the aggregation weight, and updating the global model. According to the method, the attack success rate of malicious updating can be remarkably reduced, and the method has the advantages of better robustness and safety.

The invention provides a train crosswind early warning method, a train and a storage medium, and is applied to the technical field of rail transit and the technical field of computers. The method comprises the following steps: acquiring target real-time radial wind speeds in a preset measurement range in front of a train head in two axis wave beam directions of a laser radar measured by a double-axis cross-coherent Doppler wind measurement laser radar; vector calculation is carried out on the target real-time radial wind speeds in the two axial wave beam directions, and wind speed components, perpendicular to the track direction, of the train are obtained; vector decomposition is conducted on the obtained train speed of the train running on the track, and a speed component, perpendicular to the track direction, of the train is obtained; determining the real crosswind speed of the train in the direction perpendicular to the track according to the wind speed component and the speed component of the train in the direction perpendicular to the track; and according to the real crosswind speed of the train in the direction perpendicular to the track, crosswind early warning is conducted on the train.

The invention provides a multi-level vector index construction method, a customer service scene image retrieval method and electronic device.The method comprises the steps that image vectors corresponding to customer service scene image data are obtained, the image vectors are decomposed into sub-vectors, sub-vector codebooks corresponding to the sub-vectors are constructed, and the sub-vector codebooks correspond to the customer service scene image data; obtaining an original code word index sequence corresponding to each image vector; and constructing a coarse index structure based on each original code word index sequence to obtain each cluster, and constructing a fine index structure based on each cluster to obtain a multi-level vector index structure. The problems that an existing image retrieval method is single in index structure, low in retrieval speed, difficult to balance the speed and precision and the like can be effectively solved.

The invention provides an article automatic identification method and system based on X-ray imaging and artificial intelligence, and the method comprises the following steps: obtaining dual-energy X-ray data of an article, and constructing the dual-energy X-ray data into a potential energy field model containing elevation and texture; determining a core aggregation region of each independent object in the model through gradient flow analysis, and segmenting the core aggregation region into a plurality of independent sub-regions based on a texture heterogeneity index; and carrying out vector decomposition on pixel points which generate signal aliasing due to vertical overlapping in the sub-regions by utilizing a preset material end member vector library so as to separate and reconstruct independent feature data of each material layer. And inputting each sub-region after feature reconstruction into an artificial intelligence model for category determination, and outputting a final identification result. According to the method, the technical problem of low article identification accuracy caused by the fact that a traditional AI model cannot identify composite features generated by signal superposition is solved.

The present invention provides a direction planning method, device and server for a digital simulation application, which relates to the technical field of universal development of simulation applications, including: determining the target component of a furniture model based on the interactive information of the user end, and automatically identifying the features of the target component to determine the surface feature ID of the target component; setting an absolute coordinate system, and decomposing the force vector acting on the target component into the absolute coordinate system to obtain the components of the force vector; according to the surface feature ID of the target component, correlating the preset direction planning component with the vector component, and binding the components of the force vector to the target node in the absolute coordinate system, so that the load direction is aligned with the global coordinate axis to determine the target direction vector corresponding to the interactive surface of the target component, and generating an industrial simulation application of the target furniture based on the target direction vector. The present invention can significantly improve the versatility and simulation efficiency of digital simulation applications of furniture scenes in direction planning.

The present invention provides a boundary-stripping clustering method and system based on spatial vector decomposition. The boundary-stripping clustering method comprises: step 1, input step: inputting the data set #imgabs0#; step 2, boundary-stripping step: discriminating the object type based on the distribution characteristics of the object neighborhood to strip boundary points; step 3, cluster skeleton construction step: determining the cluster skeleton based on the transitive closure relationship; step 4, boundary association step: establishing the affiliation relationship between the boundary and the cluster skeleton to complete clustering; step 5, output step: outputting the clustering results. The beneficial effect of the present invention is that the present invention not only solves the limitation of existing boundary-stripping clustering that tends to distribute uniformly and spherically, but also improves the performance of the present invention under complex distribution and high-dimensional data.

A method for calculating the levelness of a mirror plate during the turning process involves first obtaining the azimuth and levelness of the mirror plate through vector decomposition, then adjusting each support bolt to level the mirror plate, then obtaining the axis adjustment scheme after the mirror plate is leveled, and finally obtaining the rotation center adjustment scheme after obtaining the axis adjustment scheme.

The invention discloses a power amplifier linearization method based on a KAN network and a digital pre-distortion system, and belongs to the technical field of wireless communication, and the method comprises the steps: introducing Kolmogorov-Arnold Networks (KAN) in a digital pre-distortion processing process, and constructing a real value time delay KAN digital pre-distortion network and a real value time delay KAN digital pre-distortion network; for a real-value time delay KAN digital pre-distortion network, input signals at the current moment and the previous moment are considered in the input process of the network, and the memorability of a power amplifier can be correspondingly compensated; for a vector decomposition time delay KAN digital pre-distortion network, on the basis of a real value time delay KAN digital pre-distortion network, a phase recovery layer and a full connection layer are further introduced, and under the condition that the performance similar to that of a previous KAN model is kept, the complexity of the network can be simplified; compared with an original digital pre-distortion method based on a neural network, the KAN digital pre-distortion method used in the invention uses fewer network parameters and realizes higher linearization capability.