70 results about "Matrix encoding" patented technology

The invention provides a network traffic data security assessment method and system based on deep learning. The method comprises the following steps: converting original network traffic data into a graph structure data set comprising a topological structure, node attributes and time sequence behaviors; in the process, the time-space fusion input tensor is formed through the association strength between adjacent matrix and Laplacian matrix coding network entities and the fusion of time sequence characteristics extracted by time window slices. Compared with traditional flow analysis which only pays attention to a single protocol or a rate threshold value, the method achieves global relevance expression of network behaviors through graph structure modeling. Through graph structure modeling, multi-dimensional feature fusion and probabilistic risk assessment, the method can adapt to dynamic change of network topology and continuous evolution of an attack mode, so that a final assessment result is more accurate.

The invention discloses a digital twin network resource optimization method based on a multi-target genetic algorithm, which adopts a fast genetic algorithm to solve a multi-target problem so as to cope with the fast changing resource allocation requirement of a mobile ad hoc network, adopts a matrix coding mode and uses the two-dimensional data structure. According to the method, modeling is carried out on frequency domain, space domain and energy domain problems, time domain information is taken into consideration, the conflict problem of nodes of the whole network is converted into a problem model with a subnet as a unit, the decision space of an algorithm is greatly reduced, and the single operation time of iteration is shortened. The power control and frequency point allocation problem is modeled as a network transmission capacity maximization problem, so that power optimization allocation and frequency point interference minimization are unified into a multi-objective optimization problem with relevance, and the probability of power consumption and frequency point conflict can be minimized while maximization of system communication performance and frequency point utilization efficiency can be realized.

The invention relates to a security encryption sorting method and system, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: realizing key processing and data set encryption based on a preset Paillier encryption algorithm; the server performs matrix coding and comparison operation on the encrypted data set and outputs a sorting result matrix; and the two servers cooperate with each other to complete decryption of the sorting result matrix, and a plaintext sorting result is fed back to the data holder. By adopting the method, the encrypted data is sorted, the privacy leakage risk is avoided, the approximate error in the traditional technology is eliminated, and the encrypted data analysis capability which supports any data scale, is higher in accuracy and is better in calculation efficiency is realized by designing a matrix coding technology and a security comparison protocol.

The invention provides a double-input-signal virtual power system stabilizer optimization method, and relates to the technical field of power system stability control. Multi-source signals during VPSS operation are collected in real time and subjected to normalization processing, a one-dimensional time sequence signal is converted into a two-dimensional image matrix through a GASF and a GADF, and the time dependence and the dynamic phase relation of the signals are completely reserved; the physical variable frequency-power coupling is coded through a multi-channel GAF matrix; a lightweight convolutional neural network is adopted, optimization features are extracted from a GAF matrix, and VPSS control parameters are dynamically adjusted by combining a near-end strategy optimization algorithm with Lyapunov stability guarantee; introducing an online learning mechanism and an abnormal switching strategy; a closed-loop framework of signal acquisition, feature coding, mode extraction, parameter optimization, control output and abnormity recovery is adopted, system response is fed back in real time, parameters are adjusted to balance dynamic performance and steady-state efficiency of the system, and the VPSS3B is made to adapt to various operation conditions of access of the network-building type converter.

The invention discloses a feature selection method and a data reconstruction method for shoreline image feature selection. The method comprises the following steps: S1, integrating original data into a sample data set matrix X; s2, initializing a reconstruction transformation matrix W, a transformation matrix Q, an eigenvalue regression coefficient matrix A, a coordinate basis matrix B, a coding matrix E, an auxiliary matrix F and an auxiliary matrix D as unit matrixes, initializing a weight vector p and a mean vector v of a data set as unit vectors, and initializing a weight matrix P = diag (p); and S3, updating the coding matrix E, the auxiliary matrix F and the like based on the data set matrix X, the mean vector v of the data set, the transformation matrix Q, the eigenvalue regression coefficient matrix A, the coordinate basis matrix B and the weight matrix P. According to the invention, the accuracy of data reconstruction and the validity of feature selection can be improved.

The invention discloses a double-layer optimization method and system for real-time flood control scheduling of a series reservoir group. The method comprises the steps that the maximum feasible impounding boundary of each time period is calculated based on the discharge capacity and storage capacity constraint of each reservoir; constructing a space-time risk response kernel matrix, encoding lag response and attenuation intensity of flood routing, constructing a dynamic risk potential energy index in combination with real-time reservoir capacity, and solving space distribution water quantity of each reservoir by taking minimization of the index as a target under physical boundary constraint; the space distribution water quantity is used as a guide target, an evolutionary algorithm embedded with a physical consistent projection operator is adopted, and iterative optimization is carried out on hourly output flow of each reservoir. According to the method, through matrix coding of the space-time risk and introduction of manifold projection solution, the problem that upstream and downstream hydraulic space-time coupling and strong physical constraint are difficult to solve is effectively solved, and the physical consistency and peak clipping efficiency of a scheduling scheme are improved.

The invention relates to a distributed power supply-containing power distribution network fault recovery global optimization method based on ring matrix coding, and the method comprises the steps: constructing a distributed power supply model, and obtaining a distributed power supply-containing power distribution network according to the distributed power supply model; the method comprises the following steps: establishing an ordered annular matrix according to a topological structure of a power distribution network, and performing integer coding on each loop in the ordered annular matrix to obtain an annular matrix code; constructing a power distribution network fault recovery model based on the load recovery rate, the network loss and the utilization rate of the distributed power supply; and generating an initial population according to the ring matrix coding, and performing optimization solution on the power distribution network fault recovery model by adopting an intelligent optimization algorithm to obtain a final fault recovery scheme. According to the method, ring matrix coding and a one-stage fault recovery strategy are combined, all solutions generated by utilizing the ring matrix coding are feasible solutions, and the feasibility of the final solution obtained by carrying out island division on the basis is also ensured.

The invention provides a communication method and a related device. The decoding performance can be improved. The method comprises the steps that a check matrix of a target SC-LDPC code is acquired, the number of rows of the check matrix is M, the number of columns of the check matrix is N, the value of an element is 0 or 1, and the check matrix of the target SC-LDPC code comprises a second matrix composed of the ith row to the (i + (Z + 1) * m-1) th in the M rows and the 0th column to the (n-1) th column in the N columns, the second matrix is obtained by sequentially arranging (Z + 1) third matrixes with the row number of m and the column number of n according to rows, Z third matrixes exist in the (Z + 1) third matrixes, and the number K of the elements 1 included in the Z third matrixes and the number L of the elements 1 included in the second matrix meet the following relation: K / L = Z * z%; and on the basis of the check matrix coding information bit of the target SC-LDPC code, obtaining a first SC-LDPC code word.

The invention provides a communication method and device, equipment and a storage medium. The method comprises the steps that a check matrix corresponding to a target code length is determined based on a 16 * 32 basis matrix, first information obtained through coding based on the check matrix is sent, the first basis matrix supporting a first code length is associated with a second basis matrix supporting a second code length, for example, the position of a cyclic displacement matrix represented by a first element of the first basis matrix, and the position of a cyclic displacement matrix represented by a second element of the second basis matrix is associated with the position of the cyclic displacement matrix; the position of the cyclic displacement matrix represented by the first element in the first basis matrix is the same as the position of the cyclic displacement matrix represented by the first element in the second basis matrix, so that the processing complexity is reduced for the communication equipment to determine the check matrix, and the data transmission reliability of the system is improved. Furthermore, in the embodiment of the invention, in the check matrix determined based on the 16 * 32 basis matrix, the first element in the 17th column is zero, the xth element and the 16th element in the 17th column are the same non-zero elements, and x is an integer greater than 1 and less than 16, so that the coding flexibility is improved.

The invention relates to a security element (21) for a security or value document (20), the security element comprising a substrate and a combined 2D matrix code (1) that is printed on a substrate surface of the substrate and has an arrangement of modules (3) which are arranged on the substrate surface so as to define a matrix code area (2), wherein printed modules (4) of the arrangement of modules (3) are printed using colours that luminesce when illuminated with excitation light and each correspond to a colour channel of a multicolour channel system. The 2D matrix code comprises a first 2D matrix code (5), in which a first item of matrix-coded information is provided so as to be independently optically readable by means of first printed modules, and the first printed modules are printed onto the substrate in a first colour which corresponds to a first colour channel of the multicolour channel system; and a second 2D matrix code (6) which at least partially overlaps the first 2D matrix code in the region of the matrix code area (2), and in which a second item of matrix-coded information is provided so as to be readable by means of second printed modules, and the second printed modules are printed onto the substrate in a second colour which corresponds to a second colour channel of the multicolour channel system, which second colour channel differs from the first colour channel. When optically reading the combined 2D matrix code (1), the first 2D matrix code (5) can be read separately by reading the first colour channel, and the second 2D matrix code (6) can be read separately by reading the second colour channel. The invention also relates to a security or value document (20), to a method for producing a security element for a security or value document, to a method for optically reading a security element, and to a method for verifying a security or value document (20).

The embodiment of the invention provides a method and a device for recovering disk data, which are applied to a disk array, the disk array is composed of a plurality of disks, the maximum number of simultaneously invalid disks allowed by the disk array is r, and r is an integer greater than 2, and the method comprises the following steps: under the condition that the number of simultaneously invalid disks of the disk array is n, obtaining coded data from the disk array, each piece of coding data in the coding matrix respectively belongs to a diagonal check chain, a reverse diagonal check chain and a newly added check chain, and n is an integer greater than or equal to 1 and less than or equal to r; obtaining a transformation matrix of the disk array; and recovering the data in the invalid disk through the transformation matrix and the coding matrix. According to the invention, the problem that the fault-tolerant capability of the RAID6 array is relatively low is solved.

The invention provides an automatic component classification method oriented to multi-level category labels and related equipment. Comprising the following steps: preprocessing to-be-identified component data with a plurality of fields; extracting a text feature in the to-be-identified component data with the first format, and fusing the hierarchical structure information of the category label into the text feature to obtain a fused feature vector; loading a preset coding matrix based on the DECOC algorithm; the coding matrix is provided with a plurality of corresponding binary classifiers; fusing prediction probabilities of a plurality of pre-trained base classifiers based on a soft voting mechanism in each binary classifier to obtain corresponding weighted prediction probabilities so as to obtain a plurality of predicted weighted probabilities corresponding to the plurality of binary classifiers; combining the plurality of weighted prediction probabilities into a prediction code word; and determining a prediction category label group according to the prediction code word and the Hamming distance of each row in a preset coding matrix. And the types of the components can be quickly and accurately predicted.

The present application provides a communication method, an apparatus, a device, and a storage medium. The method comprises: on the basis of a 16*32 base matrix, determining a check matrix corresponding to a target code length, and sending first information obtained by encoding on the basis of the check matrix, wherein a first base matrix supporting a first code length is associated with a second base matrix supporting a second code length, for example, the position of a cyclic shift matrix represented by a first element of the first base matrix is the same as the position of a cyclic shift matrix represented by a first element in the second base matrix. For determination of a check matrix by a communication device, the processing complexity is reduced, thereby improving the data transmission reliability of a system. Further, in embodiments of the present application, in the check matrix determined on the basis of the 16*32 base matrix, the first element in the 17th column is zero, and the xth element and the 16th element in the 17th column are the same non-zero element, x being an integer greater than 1 and less than 16, thereby improving the encoding flexibility.

The invention relates to the technical field of wireless communication, and discloses a matrix coding and multi-hop link state sensing routing protocol suitable for underwater acoustic communication, which comprises the following steps of: initializing each node, and collecting own basic information of each node; each node periodically sends information messages which exist in a hop-count hierarchical coding matrix form and are used for information exchange to neighbor nodes so as to preliminarily construct a network topology view of each node; integrating and perceiving the network topology view of each node so as to form a network topology database of each node; finding a routing path between the source node and the destination node according to a preset rule for dynamic query; carrying out data transmission based on the queried routing path, and monitoring a link state and a state of a one-hop neighbor node of the node to be subjected to data transmission in real time in a data transmission process; judging whether the receiving node receives the data sent by the sending node or not; and judging whether the sending node receives the confirmation message. According to the invention, the convergence of the routing table can be quickly completed.

The invention discloses an image segmentation method and device for a fuel particle layered micro-nano structure. The method comprises the steps of data preprocessing, wherein 3D CT data are converted into 2D Cartesian slices, polar coordinate images are obtained, and strips are divided by means of particle radial symmetry; feature extraction: constructing a main network by U-Net, and adding a position coding module (strengthening angle position modeling) and a disturbance module (improving generalization ability); the multi-class segmentation is divided into boundary dichotomy subtasks, and boundary perception weighting loss is combined with weighted binary cross entropy training. And feature fusion: defining a structure priori matrix coding coating sequence, and training the weight by classifying cross entropy through a conditional random field fusion result (ensuring topological consistency between layers). And finally, splicing strip blocks, and converting polar coordinates into a Cartesian image. The method is applied to automatic nuclear fuel detection and radiation imaging, multi-layer accurate segmentation is achieved, and the detection efficiency and precision are improved.

The application discloses a waiting time constraint batch processing pipeline crane scheduling method based on an improved slime algorithm and belongs to the field of intelligent scheduling of industrial production. The method comprises the following steps: a scheduling model is constructed, the waiting time fuzzy interval is defined by a trapezoidal fuzzy number, and the maximum completion time is minimized as the target; the waiting time uncertainty is accurately characterized by the fuzzy interval, and the fluctuation range of the waiting time in actual production is completely covered; based on the model, a hybrid improved slime algorithm is used for solving; a matrix coding and One-Hot mapping are adopted to adapt to the discrete scheduling scene; a good point set initialization method is introduced to construct an initial population, and multiple scheduling schemes are obtained; and a double tabu table strategy is combined to optimize the local search efficiency and convergence stability. The application is superior to traditional genetic algorithms, discrete slime algorithms and other comparative algorithms in terms of solution quality, convergence speed and robustness, can efficiently adapt to the crane scheduling demand of process-type industries, and provides a stable and reliable solution for multi-crane collaborative scheduling in a complex uncertain environment.

The invention provides an interference communication integrated signal design method based on MIMO-OCDM (Multiple Input Multiple Output-Optical Code Division Multiplexing). The implementation steps are as follows: constructing an application scene of an interference communication integrated signal and a binary communication data matrix; performing OCDM on the binary communication data matrix, and obtaining an MIMO-OCDM communication signal and a frequency shift interference signal; and finally, an MIMO-OCDM interference communication integrated signal is obtained. According to the method, Hadamard matrix coding is carried out on the time domain signal matrix obtained after discrete Fresnel inverse transformation is carried out on the frequency domain QPSK symbol matrix so as to obtain the MIMO-OCDM communication signal matrix, it is guaranteed that multiple time domain signals have the high anti-interference capacity, interference of signals between MIMO transmitting channels is reduced, and the communication performance of the MIMO-OCDM communication signal matrix is improved. Moreover, a non-partner needs to obtain the interference communication integrated signals transmitted by all MIMO transmission channels and correctly analyze and recombine the interference communication integrated signals to obtain the communication data, thereby improving the anti-interference difficulty of the radar of the opposite side.

The invention relates to a security element (21) for a security or valuable document (20) and a method for manufacturing the security element (21), which comprises a substrate and a combined 2D matrix code (1), wherein the combined 2D matrix code (1) is: printed on a substrate surface; comprises an arrangement of modules formed with printed modules printed using inks that are optically readable under illumination with light; comprises a first 2D matrix code (2) in which a first matrix-coded information is provided by means of first printed modules, and the first printed modules are printed on the substrate with a first ink that is optically readable under illumination with light from a first spectral range; and comprises a second 2D matrix code (3).which is formed in the area of ​​the matrix code surface (2) and is formed at least partially overlapping with the first 2D matrix code, and in which a second matrix-coded information is provided by means of second printed modules, and the second printed modules are printed on the substrate with a second color, which can be optically read under illumination with light from a second spectral range that differs from the first spectral range. When optically reading the combined 2D matrix code (1), the first matrix-coded information can be read by reading the first 2D matrix code (2) under illumination with light from the first spectral range, and the second matrix-coded information can be read by reading the second 2D matrix code (3) under illumination with light from the second spectral range. Furthermore, a security or valuable document (20),a method for optically reading a security element and a method for checking a security or value document (20) as well as a method for optically reading are provided.

This invention discloses an automatic data repair method for missing traffic loop data based on latent feature learning. The method involves matrix encoding of traffic flow data collected by loops to obtain a traffic flow parameter value matrix. By setting latent feature matrices, temporal latent feature matrices, and loop latent feature matrices, and establishing a traffic flow parameter value calculation model, the optimal temporal latent feature matrix and loop latent feature matrix suitable for the traffic flow data are solved using the stochastic gradient descent method. Multiplying these matrices yields the missing traffic flow parameter values, thereby improving data quality and effectively enhancing the accuracy of various traffic statistics and model precision. This invention is the first to propose a traffic flow matrix encoding method, the first to propose traffic temporal latent feature learning and loop latent feature learning methods, and the first to apply them to repair missing traffic flow parameter data.

Embodiments are disclosed for a matrix coded stereo signal with periphonic elements. A mixing matrix, suitable for processing a multi-channel audio input signal, is constructed so that the resulting multi-channel output signal contains the same audio elements from the input signal, wherein the spatial relationships between audio elements, as defined by panning rules associated with the input signal format, are preserved in the output signal, as defined by matrix encoding rules associated with the output signal format. The choice of the coefficients of the mixing matrix is governed by a phase-preference rule that is used to determine the appropriate phase to apply to each input signal channel.

The embodiment of the invention provides a communication method and device. The decoding end determines a first coding matrix; the first coding matrix is determined according to at least one received first signal and / or at least one acquired interference signal between a first time for determining the first coding matrix and a second time for determining the second coding matrix, the first signal comprises the interference signal, and the second coding matrix is a coding matrix determined before the first coding matrix. And the decoding end sends first information to the coding end, wherein the first information is used for indicating the first coding matrix. And the encoding end encodes a signal to be sent to the decoding end based on the first encoding matrix. According to the method, the decoding end determines the coding matrix once every a period of time and determines the coding matrix based on the interference signal, the coding matrix adopted by the coding end is variable and changes along with the interference of the surrounding environment, and compared with coding by using a factory fixed coding matrix, the decoding end can estimate the interference signal more accurately, and the coding efficiency is improved. And communication quality is improved.

The invention discloses an RS-485 bus-based equipment unique identification code acquisition method, which is suitable for communication between a host and a plurality of slaves, and each slave comprises an equipment memory module used for storing an identification code with uniqueness of equipment; according to the acquisition method, matrix coding is used for communicating and interacting with a host and a slave, accurate positioning is realized by adopting an iterative combination mode according to conflict dynamic screening, and thus the unique identification code for identifying the equipment is acquired. By adopting the acquisition method provided by the invention, the communication complexity can be remarkably reduced through matrix coding, staged verification and an iterative screening mechanism, so that the automatic identification requirement of the equipment is met, and the acquisition method can be widely applied to large-scale equipment networking scenes such as intelligent power grids and industrial automation; and bottom support is provided for equipment management, fault location and security authentication.

A user equipment (UE) reports channel state information (CSI) to a base station in multiple-input multiple-output (MIMO) transmissions. The UE constructs a CSI matrix based on a CSI reference signal (CSI-RS) received from the base station. The CSI matrix is at least three-dimensional in a transmit (Tx) antenna domain, a frequency domain, and a time domain. The UE transforms the CSI matrix into a transformed matrix in at least a Tx beam domain, a delay domain, and a Doppler domain. The UE encodes the transformed CSI matrix into a one-dimensional feature vector using a multi-layered neural network, and sends the one-dimensional feature vector to the base station.

The invention relates to a security element (21) for a security or valuable document (20) comprising a substrate and a combined 2D matrix code (1) printed on a substrate surface of the substrate with an arrangement of modules (3) which are arranged on the substrate surface to cover a matrix code area (2), wherein printed modules (4) of the arrangement of modules (3) are printed using colors which luminesce under illumination with excitation light and each correspond to a color channel of a multicolor color channel system.The 2D matrix code comprises a first 2D matrix code (5) in which a first matrix-coded piece of information is provided for independent optical reading by means of first printed modules, and the first printed modules are printed on the substrate with a first color which corresponds to a first color channel of the multicolor color channel system; and a second 2D matrix code (6) which is formed in the area of ​​the matrix code surface (2) at least partially overlapping with the first 2D matrix code and in which a second matrix-coded piece of information is provided for reading by means of second printed modules, and the second printed modules are printed on the substrate with a second color which corresponds to a second color channel of the multicolor color channel system, which is different from the first color channel.When optically reading the combined 2D matrix code (1), the first 2D matrix code (5) can be read separately by reading the first color channel and the second 2D matrix code (6) can be read separately by reading the second color channel. Furthermore, a security or valuable document (20), a method for producing a security element for a security or valuable document, a method for optically reading a security element, and a method for verifying a security or valuable document (20) are provided.

The present invention relates to the technical field of power battery disassembly and recycling, and discloses a method for dynamically determining the disassembly path of retired batteries based on depth search. A disassembly path is constructed based on the battery disassembly process steps, and a graph matrix encoding is constructed according to the disassembly path; the machine state is obtained, and the machine state and the graph matrix encoding are input into the pre-constructed reinforcement learning model to obtain the disassembly weight; the first optimized disassembly path is obtained by combining the disassembly weight with depth search, and disassembly is carried out according to the first optimized disassembly path, and battery information is obtained. According to the battery pack information, the second optimized disassembly path is obtained by combining the reinforcement learning model and depth search again, and the machine is allocated according to the second optimized disassembly path to complete the disassembly of the retired battery, solving the problem that the existing disassembly path cannot be adjusted in real time.

The invention discloses an injection molding defect knowledge graph construction method and system based on deep learning and a medium, and relates to the technical field of data processing, and the method comprises the steps: obtaining real-time sensor data and mold flow simulation data of an injection molding machine, constructing an initial graph, inputting a space-time diagram convolution network to calculate hidden association between nodes, and outputting an association probability matrix; generating a defect texture image based on the matrix; and coding the matrix into predicate vectors, performing momentum contrast learning on the predicate vectors and image feature vectors, constructing a cross-modal shared semantic space, performing fine tuning by using real defect samples, and outputting an interpretable defect diagnosis model. The technical problems that an existing injection molding defect diagnosis method mostly focuses on data analysis of a single mode, complex relations between process parameters and defects are difficult to capture, and model interpretability is poor are solved, and sample enhancement and complex relation capture are achieved through deep learning and the generative adversarial network. And the interpretability and practicability of the model are improved.

A dynamic coding construction and low-complexity coding method and apparatus for error correction and encryption fusion, the method comprising: continuously receiving a sequence of plaintext information sent by a source, and constructing a corresponding dynamic sparse random parity check matrix H for each plaintext information. i For each H i Perform matrix block processing to obtain the H i The corresponding multiple submatrices and a double diagonal matrix T i According to T i Calculate the intermediate matrix φ from the inverse matrix and submatrix i For every K intermediate matrices φ obtained i Then, the grouping inverse algorithm is used to calculate each φ. i The corresponding inverse matrix K≥2, for each plaintext m i Based on the corresponding dynamic matrix encoding, m is obtained. i The corresponding ciphertext P is sent. i The encoding matrix is ​​constructed by combining a double diagonal matrix structure and a dynamic sparse random parity check matrix block processing method. The encoding process uses a group inversion algorithm, which reduces the complexity of obtaining the encoding matrix and the computational complexity of encoding, thus greatly improving the encoding efficiency.

The present invention relates to the technical field of resource recycling, and discloses a disassembly scheduling method for retired batteries based on reinforcement learning. The method includes: obtaining the machine state of the disassembly machine, constructing a graph matrix encoding based on the battery disassembly process steps, constructing an MGCN unit for extracting the connectivity and topological structure between different parts of the battery based on the Conv2d k1 block and the Conv2d k3 block, and constructing a multi-graph convolutional network according to the MGCN unit; inputting the machine state and the graph matrix encoding into the multi-graph convolutional network to obtain disassembly weights; obtaining an optimized disassembly path by combining depth search according to the disassembly weights, and completing the disassembly of the retired battery according to the optimized disassembly path, which solves the problem of low efficiency of existing models such as existing reinforcement learning when applied to battery disassembly scheduling.

The present disclosure relates to a data processing method, device, medium, equipment and product for secure computation. The secure computation is used for a first participant and a second participant to cooperatively compute the product of a target matrix and a target column vector. The method comprises the following steps: encoding the target matrix into N polynomials; generating a second ciphertext polynomial according to the N polynomials and a first ciphertext polynomial sent by the second participant; generating an N-dimensional random column vector, performing a masking process on the second ciphertext polynomial by using the random column vector, and sending the obtained masking polynomial after the masking process to the second participant; taking the random column vector as a second shard of the product, and performing a target data processing task based on the second shard. In this way, the product of the matrix and the vector can be securely computed by using the properties of the operator ring, and the present scheme can be applied to the secure computation of the product of the matrix and the vector on the ring of any modulus. In addition, the data security can be protected and the applicability can be improved in the model training scene by the present scheme.