25 results about "Full matrix" patented technology

A communication method and a communication apparatus in which a device may perform encoding or decoding based on an LDPC matrix, the LDPC matrix being determined based on an LDPC basis matrix, Zc, a, and b, a first region of the basis matrix including L types of elements, 2 < = L < = 2a * b, a and b not being 1 at the same time, Zc, a, and b being used to lift the elements in the first region to a (a * Zc) * (b * Zc) matrix, the matrix including (a * b) (Zc * Zc) matrices, each Zc * Zc matrix is an all-zero matrix or a cyclic shift matrix, the first sub-matrixes of the L elements are different, and the first sub-matrixes are matrixes formed by replacing the cyclic shift matrixes in (a * b) (Zc * Zc) matrixes of the matrix with the lifted corresponding elements with Zc * Zc unit matrixes. According to the method, the decoding performance of the LDPC code can be improved.

The invention relates to the technical field of artificial intelligence calculation models and intelligent decision control, and discloses a system based on an AI multi-dimensional model and an intelligent decision method, and the system comprises a feature perception module, a topology reconstruction engine, a feedback scheduling unit and a transformation processing unit. The feature sensing module extracts feature vectors of all components of the high-dimensional heterogeneous data stream, calculates a gradient fusion value and determines an active feature dimension; a topology reconstruction engine generates a dynamic mask based on the active feature dimensions, and performs edge pruning on the initial topological graph to construct a task-sensitive topological structure; the feedback scheduling unit uses a reinforcement learning operator to adjust a pruning threshold value and updates a dynamic mask coverage rate; and the conversion processing unit executes weighted mapping operation on the basis of the updated dynamic masks, full matrix operation is converted into sparse topological operation through feature guide logic pruning, and invalid electrical digital processing overhead is reduced.

The application discloses a kind of fusion social network graph and hypergraph information recommendation system matrix filling method, comprising the following steps: S1, using generative model modeling generation social network graph information and hypergraph information;S2, the generation and input of incomplete recommendation matrix;S3, through three-stage algorithm, using social network hypergraph information and incomplete matrix information are carried out matrix filling, output the recommendation matrix after completion, and then complete recommendation task;S4, collect user feedback information, optimize recommendation algorithm: according to the feedback of user, improve social network information and then optimize recommendation algorithm.Compared with prior art, the present application uses mixed graph and more complex hypergraph for modeling, can better make full use of complex social network data, on this basis, through three-stage algorithm, using social network hypergraph information auxiliary matrix filling, can carry out more accurate matrix reply, and then enable recommendation system to realize more efficient recommendation.

The application relates to the field of ultrasonic nondestructive testing, and discloses a downhole thread digital modeling method and system. The method comprises the following steps: firstly, wrapping a thread area with an adaptive coupling module; secondly, fixing an ultrasonic phased array probe on the adaptive coupling module, performing a compound motion, and acquiring full matrix data; then, performing imaging processing on the full matrix data to obtain a two-dimensional image; then, extracting a thread three-dimensional feature point cloud based on the two-dimensional image; and finally, reconstructing a three-dimensional digital model of the thread based on the obtained three-dimensional feature point cloud. The method has the sound beam deflection and focusing capacity of the phased array technology, can better adapt to threads of different specifications and surface states in combination with flexible coupling, has strong universality, and in addition, the method can not only judge whether a defect exists, but also accurately reconstruct a three-dimensional form of the defect, thereby providing unprecedented data support for safety evaluation and service life prediction.

The invention provides a self-adaptive ultrasonic detection and defect identification method for a composite material, and belongs to the technical field of ultrasonic detection.The self-adaptive ultrasonic detection and defect identification method for the composite material comprises the specific steps that firstly, preparation is conducted before detection; 2, collecting full-matrix data of the one-way plate; 3, calculating a core elastic constant of the composite board; 4, deducing the velocity distribution of each layer group of the composite board; 5, collecting full-matrix data of the composite board; step 6, adaptively correcting imaging delay; and 7, performing full-focusing imaging and judgment on the defects of the composite board. According to the method, defect subtle features which are difficult to recognize by human eyes in the full-focus imaging picture can be automatically mined through the deep learning model, the problems of missing detection and false detection caused by subjective judgment deviation are effectively avoided, rapid and accurate classification and positioning of various defects can be realized through the deep learning model, the detection period is greatly shortened, and the detection efficiency is improved. And reliable technical support is provided for high efficiency and precision of ultrasonic detection of the composite material.

The invention relates to the technical field of power control, in particular to a full-matrix power distribution module, unit and equipment. The full-matrix power distribution module enables input control parts of the relay units in the same row to share lines among different relay groups; meanwhile, the power parts of the adjacent relays in each relay group are connected in a back-to-back manner, so that the power output parts of the adjacent relays share lines, multi-path power distribution full-matrix output is realized, and the problem of low density of traditional full-matrix power distribution is solved. Meanwhile, at the output end, the bus output port of each relay group is provided with an output external wire in a step mode, so that front installation and assembly are facilitated.

This invention relates to the field of artificial intelligence computing models and intelligent decision control technology, and discloses a system and intelligent decision-making method based on an AI multi-dimensional model, including a feature perception module, a topology reconstruction engine, a feedback scheduling unit, and a transformation processing unit. The feature perception module extracts feature vectors of each component of a high-dimensional heterogeneous data stream, calculates the fusion value of the gradient, and determines the active feature dimension. The topology reconstruction engine generates a dynamic mask based on the active feature dimension and performs edge pruning on the initial topology graph to construct a task-sensitive topology structure. The feedback scheduling unit uses reinforcement learning operators to adjust the pruning threshold and update the dynamic mask coverage. The transformation processing unit performs weighted mapping operations based on the updated dynamic mask. This invention transforms full matrix operations into sparse topology operations through feature-guided logical pruning, reducing the overhead of invalid digital processing.

The utility model discloses a full-matrix power distribution topological structure based on charging piles, which belongs to the technical field of circuit electronics and comprises a plurality of power module groups, a plurality of power distribution units and a plurality of charging terminals. The power module group comprises a combination of one or more power modules, and the technical problem that the cost is high due to the fact that an existing power distribution topological structure needs to be provided with a plurality of power distribution units and a plurality of contactors is solved.

The invention relates to a dynamic sparse ultrasonic full-focusing method for a large-wall-thickness structural member, belongs to the field of nondestructive testing, and solves the problems of low full-focusing imaging efficiency, insufficient deep defect detection sensitivity and unbalanced imaging quality at different depths caused by a fixed sparse rate during ultrasonic testing of the large-wall-thickness structural member (the thickness is greater than or equal to 300mm) in the prior art. The method comprises the following steps: S1, acquiring medium sound velocity and total thickness of a detected component and determining array parameters of the detected component; s2, collecting a full-matrix signal covering a detection depth range; s3, establishing an acoustic beam diffusion model, and calculating acoustic beam diffusion widths at different depths; s4, dynamically distributing the array element sparse rate of each depth; and S5, selecting an effective array element according to the sparse rate, calculating the delay time from the effective array element to the focus point, and reconstructing and imaging to obtain a full-focus detection image covering the whole detection depth. Efficient and accurate nondestructive testing of large-wall-thickness structural members such as pressure containers and heavy pipelines is achieved, and imaging quality and detection efficiency are both considered.

A system and method for an adaptive full focus method (TFM) includes obtaining full matrix capture (FMC) data of a region of interest of an article, the region of interest being made of a material; for each unit of a set of units in the article, evaluating the FMC data to obtain a parameter for relating to a material property of the material; associating the parameters of each cell in the set of cells with a state model of the material to determine an anisotropic state of each cell in the set of cells in the article; generating a distribution parameter model of the material anisotropy of the product by using the anisotropy state of each unit in the group of units in the product; and operating the beamformer with the distributed parameter model using the TFM to image the region of interest.

A communication method and a communication apparatus. In the method, a device may perform encoding or decoding on the basis of an LDPC matrix, wherein the LDPC matrix is determined on the basis of an LDPC base matrix, Zc, a, and b; a first region of the base matrix comprises L types of elements, wherein 2≤L≤2a*b, and a and b are not both 1 at the same time; Zc, a, and b are used for lifting each element in the first region into an (a*Zc)*(b*Zc) matrix, wherein the matrix comprises (a*b)(Zc*Zc) matrices, and each Zc*Zc matrix is an all-0 matrix or a cyclic shift matrix; first sub-matrices corresponding to the L types of elements are all different; and each first sub-matrix is a matrix formed by replacing, with an identity matrix of Zc*Zc, a cyclic shift matrix within the (a*b)(Zc*Zc) matrices of the matrix lifted by a corresponding element. The method can improve the decoding performance of LDPC codes.

Mechanisms are provided for fault-tolerant computing. The mechanisms solve, by a controller computing device in a controller-worker computer architecture, a sparse algebraic linear system of equations with incomplete matrix-vector products. The solving includes: A set of worker computing devices computing matrix-vector products Ax=b, wherein A is a sparse matrix, x is a vector being multiplied, and b is a resultant vector; a controller determining that a worker has not returned a result of its computation within a threshold length of time; and in response to the determination, assuming, by the controller computing device, that the result of the worker computing device computation is zero and continuing the computing of a solution to the sparse algebraic linear system. The controller computing device provides, to a subsequent computing operation, the solution to the sparse algebraic linear system of equations as a basis for performing the subsequent computing operation.

This invention relates to a secure computation method and system for nonlinear functions based on unintentional shuffling and splitting outsourcing, belonging to the field of computer software technology. The method includes: generating dynamic permutation parameters for inference requests via a client; performing unintentional shuffling on the ciphertext through two mutually distrustful servers based on the dynamic permutation parameters; splitting the unintentionally shuffled data into a first part and a second part, with each server reconstructing the first and second parts respectively to obtain partial plaintext; calculating a nonlinear function using the partial plaintext and merging the results; and performing inverse unintentional shuffling on the merged results to obtain the final nonlinear layer output. This invention transforms expensive online secure matrix multiplication into low-cost vector addition and single-round communication operations, achieving "one-time pad" dynamic permutation at extremely low cost, fundamentally resisting statistical inference attacks.

The invention discloses a full-matrix signal conversion method for driving apodization plane wave signals to a synthetic emission aperture (STA) by a self-supervised convolutional neural network (CNN) for high-quality two-dimensional ultrasonic imaging, an ultrasonic imaging system and a medium. The method comprises the following steps: designing a self-supervised learning CNN model based on a linear acoustic superposition principle; training the CNN model by using the simulation STA full matrix signal; an optimized binary weight is extracted from a CNN model encoder convolutional layer to serve as emission apodization, and plane wave signals are collected; inputting the apodization plane wave signal into a CNN model decoder to convert and generate an STA full-matrix signal; and performing beam forming based on the generated STA full-matrix signal so as to reconstruct a high-quality B ultrasonic image. According to the method, apodization plane wave signals can be well converted into STA full-matrix signals so as to carry out high-quality two-dimensional ultrasonic imaging, and the problem that the plane wave imaging quality is poor can be solved to a certain extent.

The present application relates to the technical field of casting billet quality evaluation, and particularly relates to a quality evaluation method and system for a casting billet forming process, which comprises the following steps: emitting an ultrasonic pulse to a detection area of the casting billet, receiving a full-matrix echo signal, and the full-matrix echo signal comprising a primary echo signal and a second harmonic signal; performing variational mode decomposition on the primary echo signal to separate a grain boundary scattering signal and a linear defect echo signal generated by a coarse grain structure of the casting billet; constructing a scattering texture atlas based on the grain boundary scattering signal and constructing a linear feature defect set based on the linear defect echo signal; performing pulse inversion processing on the second harmonic signal to construct a nonlinear response atlas; performing spatial registration on the scattering texture atlas, the linear feature defect set and the nonlinear response atlas to construct a multi-dimensional fusion feature vector; and obtaining a quality evaluation result of the casting billet forming based on the spatial distribution characteristics of the multi-dimensional fusion feature vector in the detection area. The present application effectively improves the accuracy of quality evaluation of the casting billet forming process.

The application provides an ultrasonic-based forging flaw detection method and system, full matrix data collected by an ultrasonic array scanning along a preset path on a forging surface is acquired; symmetric mode components and antisymmetric mode components of Lamb waves are separated; a signal reflected from a known geometric structure of the forging is used as a reference to establish a propagation model in a frequency-wavenumber domain, and a frequency-dependent attenuation coefficient distribution of a non-uniform sound velocity distribution inside the forging is determined; for any pixel point of an imaging area to be imaged, a focusing operator for compensating signal propagation time and amplitude is constructed for each transceiving element pair; the focusing operator is used to perform synthetic aperture focusing on the symmetric mode components and the antisymmetric mode components respectively, and symmetric mode focusing images and antisymmetric mode focusing images are obtained; a fusion criterion is determined based on energy and phase information of the pixel points in the two images, the symmetric mode focusing images and the antisymmetric mode focusing images are weighted and fused, and an internal defect image of the forging is generated.

This invention discloses a method and system for identifying the corrosion state of wind turbine bolt materials. The method includes: acquiring raw signals using a ring ultrasonic transducer array in full-matrix acquisition mode; obtaining focused echo signals through full-focusing post-processing; performing wavelet packet decomposition and soft-threshold denoising on the focused echo signals to extract the energy entropy of each sub-band; concatenating the raw time-domain signal, sub-band reconstructed signal, and energy entropy into a multi-channel feature sequence; inputting the sequence into a deep convolutional neural network combined with a channel attention mechanism to extract corrosion feature semantic vectors; acquiring load and environmental data and mapping them to external feature vectors, fusing them with the corrosion feature semantic vectors through tensor product fusion to obtain a dynamic embedding vector; constructing the dynamic embedding vectors of the same bolt at different times as an input sequence, and outputting the corrosion level, residual strength ratio, and remaining life in parallel through a bidirectional long short-term memory network. This invention can improve the accuracy and multi-dimensional quantitative assessment capability of identifying the corrosion state of wind turbine bolts under complex and harsh working conditions.

This invention relates to the field of nondestructive testing (NDT) technology, specifically disclosing an ultrasonic NDT method based on full-matrix data transfer and phase coherence, used to detect internal damage defects in planar multilayer structures in equipment such as pressure vessels. This method utilizes the time-domain echo signal received by an ultrasonic phased array body wave probe, combined with frequency-domain phase shift transfer of the full-matrix dataset to achieve rapid extrapolation of the wave field, improving the algorithm's time efficiency by avoiding interface refraction operations of the time-domain signal. Based on the explosion reflection model, the time-domain dataset of the target region is extracted, and combined with a phase-coherence weighted full-focusing imaging algorithm, accurate localization of internal damage within the target layer of the planar multilayer structure is achieved. Experimental results show that this invention has high accuracy and resolution in defect localization imaging of planar multilayer structures, and is suitable for the inspection and maintenance of planar multilayer walls in equipment such as pressure vessels.

The application provides a method for solving dynamic voltage drop of a chip-package based on LU decomposition combined with an SMW formula. The chip-package power supply network is regarded as a composite system connected by an interconnection node Pad between a package part and a plurality of chip sub-networks. A chip-package partition solving strategy is introduced. According to the chip and package characteristics, a large-scale sparse matrix of the composite system is disassembled into a structural model with a plurality of diagonal sub-blocks, so as to convert the chip-package power supply network equation into a block structure, and avoid the inefficient redundancy of full matrix repeated decomposition by focusing on necessary coupling calculation. Meanwhile, the SMW formula is used for low-rank incremental updating of the package network influence, so as to quickly update the influence introduced by the package and quickly correct the solution vector. The calculation task is split into a plurality of sub-problems for synchronous execution, so that the plurality of sub-blocks are solved synchronously. The application can provide a high-precision and high-efficiency solution for dynamic voltage drop analysis of a large-scale chip package power supply network.

The application provides a kind of casting billet online thermal state tracking method, comprising the following steps: according to the rule of block division, the asymmetric profiled blank is divided into area block, and the casting billet center area block and the casting billet surface area block are obtained respectively;The structured coarse grid division is carried out to the casting billet center area block, and the structured fine grid division is carried out to the casting billet surface area block, to obtain a plurality of structured grids, and the grid nodes of the structured grid are indexed and described, and the grid data of each structured block and the index are stored correspondingly;The heat transfer equation coefficient matrix and the solidification latent heat release source term of the grid unit of each structured block are calculated;The temperature field of the grid unit in the structured block is calculated by coupling the incomplete matrix LU decomposition and SIP algorithm in turn. By using the application, the problems of the prior art, such as unsuitable for asymmetric profiled blank and slab dual-purpose casting machine, reduced system reliability and time step limitation, can be solved.

A method of recovering transmitted symbols. The method receiving a signal comprising a codeword, the signal having been affected by channel effects including distortion and noise. A solution space for recovering the transmitted symbols is identified, including finding a sampling center using a full matrix W for minimum mean square error. A vector v′ is generated to generate noise. The vector has a variance ρσn2 where σn2 is a factor that defines power of added noise and where ρ is a scaling parameter that controls a size of a noise sampleable space. The vector v′ is applied to the solution space. Original samples are gathered from the solution space, which includes noise from the vector v′ and symbol information from the received signal to find probabilities for symbols. Using the probabilities, the symbols are recovered.

This invention discloses a high signal-to-noise ratio (SNR) ultrasonic imaging method for internal defects in materials. By utilizing the spatial coherence of the received signals from each channel in the full matrix data and modifying the delay rule of the algorithm based on the surface configuration characteristics of PBX (partially filled composite material), high SNR imaging of crack defects with different orientations in PBXs with curved surface configurations is achieved. The quality of the reconstructed images is compared and evaluated. The results show that, compared with traditional linear synthetic focusing imaging, the algorithm proposed in this invention improves the SNR of ultrasonic reconstructed images of crack defects in particulate-filled composite material PBX by more than 10 dB, significantly improving the detection capability of crack defects while maintaining imaging efficiency.

This invention relates to a processing method and apparatus for a dense encoder based on a large language model. The method includes: selecting a large language model that has been pre-trained and fine-tuned for NLP tasks and implemented based on a pure decoder architecture as the target model; obtaining a bidirectional encoder by solidifying the causal mask matrix used by the target model's decoder during inference into an all-one matrix; and forming a dense encoder by sequentially connecting the target model's embedding encoding module and multiple bidirectional encoders; performing a first-stage fine-tuning of the dense encoder using a masked word prediction task; performing a second-stage fine-tuning of the dense encoder using an unsupervised contrastive learning mechanism; after fine-tuning, constructing a document vector library for a user-specified target document library using the dense encoder; and providing retrieval services for the target document library based on the document vector library and the dense encoder. Using the dense encoder of this invention to process text retrieval tasks can improve retrieval accuracy.

The invention discloses a power failure positioning method and device based on multi-weight verification and a readable medium, and the method comprises the steps: S1, constructing an m * 5-dimensional parameter matrix R for each feeder line every day; s2, initializing an m * 5-dimensional fault matrix X into an all-0 matrix, and updating the assignment of the fault matrix X according to the received transformer fault alarm signal of each feeder line; s3, multiplying the fault matrix X and the parameter matrix R by bits to obtain a collection result matrix Y, and carrying out fault hierarchy collection according to a preset trigger condition; s4, judging the power failure range and type according to the sum s of the first columns of the fault matrix X and the collection result matrix Y; and S5, repeatedly executing the step S3 and the step S4 at intervals, and determining the power failure type if multiple continuous results are consistent. According to the invention, the problems of insufficient positioning precision, lack of correlation analysis and poor anti-interference performance in the existing power failure detection technology are effectively solved.

The invention provides a forge piece flaw detection method and system based on ultrasonic waves. Full-matrix data collected by scanning of an ultrasonic array on the surface of a forge piece along a preset path is obtained; separating a symmetric mode component and an antisymmetric mode component of the Lamb wave; the method comprises the following steps: establishing a propagation model in a frequency-wavenumber domain by taking a signal reflected from a known geometric structure of a forging as a reference, and determining non-uniform sound velocity distribution in the forging and attenuation coefficient distribution related to frequency; a focusing operator used for compensating signal propagation time and amplitude is constructed for each transmit-receive array element pair for any pixel point of the to-be-imaged area; performing synthetic aperture focusing on the symmetric mode component and the antisymmetric mode component by using a focusing operator to obtain a symmetric mode focused image and an antisymmetric mode focused image; and determining a fusion criterion according to energy and phase information based on pixel points in the two images, and carrying out weighted fusion on the symmetric mode focusing image and the antisymmetric mode focusing image to generate an internal defect image of the forge piece.