Patents
Literature
Patsnap Eureka AI that helps you search prior art, draft patents, and assess FTO risks, powered by patent and scientific literature data.

70 results about "Z eigenvalue" patented technology

Smart community-oriented multi-modal sensor data real-time fusion processing method

PendingCN120873978ABiological modelsFractional Brownian motionAlgorithm
The invention relates to the technical field of data processing, in particular to a multi-modal sensor data real-time fusion processing method for a smart community. According to the method, a sensor network topological graph is constructed, a connection weight is optimized, distributed clock synchronization is realized by using a graph Laplacian matrix, and clock drift prediction and compensation are performed in combination with a fractional Brownian motion model; performing wavelet transform decomposition on the sensor data after time sequence alignment, calculating each scale Hurst index, predicting a load trend through a fractal prediction model, and outputting an optimal resource allocation scheme through hybrid evolution calculation; the method comprises the following steps: constructing multi-modal sensor data into a graph structure, extracting node features by using a graph convolutional neural network, obtaining global feature representation by using a self-attention mechanism, and performing anomaly detection classification in combination with a resource utilization rate and a prediction error; an anomaly detection feedback mechanism is established, and Laplacian matrix eigenvalues and weight parameters are dynamically adjusted; the real-time performance, the accuracy and the robustness of data fusion processing are improved.
Owner:ZHEJIANG YUMAI TECH

Level of detail via numerically stable eigenvalue technology

Methods, systems and apparatuses may provide for technology that determines a major axis length of an ellipse in texture space based on a transformation matrix associated with a circle in screen space, inverts the major axis length, determines a product between the inverted major axis length and a determinant of the transformation matrix, and determines a minor axis length based on a modulus of the product between the inverted major axis length and the determinant of the transformation matrix. In one example, the technology selects a level of detail in the texture space based on the major axis length, the minor axis length, a direction of the major axis length, and a direction of the minor axis length.
Owner:INTEL CORP

Multi-converter parameter collaborative optimization method for improving small interference stability of active power distribution network

The invention discloses a multi-converter parameter collaborative optimization method for improving the small interference stability of an active power distribution network, and belongs to the technical field of power systems, and the method comprises the steps: building a differential algebraic equation model, and carrying out the linearization of a steady-state operation point, and obtaining a state space model; and a control parameter optimization problem taking the minimization of the spectrum intercept construction as a target is defined, and control parameters are a GFL converter phase-locked loop parameter and a GFM converter virtual damping parameter. First-order sensitivity and second-order sensitivity of eigenvalues to control parameters are calculated based on a matrix perturbation theory and are converted into local quadratic constraint sub-problems. And performing convexity solution by adopting a positive semidefinite relaxation method, adjusting the step length in combination with a line search mechanism, iteratively updating control parameters, and outputting an optimization result after a convergence condition is met. According to the method, the spectral intercept is reduced through multi-converter parameter collaborative optimization, and the weak stability characteristic is improved; the convergence and solvability are improved by utilizing matrix perturbation and positive semidefinite relaxation; and the stability is enhanced and optimized by adopting line search step length adjustment.
Owner:GUIZHOU POWER GRID CO LTD

Flow field evolution method and device based on quantum calculation and vortex representation

The invention provides a flow field evolution method and device based on quantum calculation and vortex characterization, and the method comprises the steps: defining a complex scalar field, constructing a Hermi matrix based on an initial velocity field, and converting a vortex thread positioning problem into a quantum ground state solving problem; solving the minimum characteristic value of the matrix through a quantum algorithm, and outputting an initial vortex quantum state; according to a fluid dynamic control equation, constructing Hamiltonian through discretization processing; constructing a Hamiltonian quantum circuit, and driving the initial vortex quantum state to evolve to a target quantum state; and measuring the evolved quantum state, obtaining a discrete numerical value of the complex scalar field on the central point of the grid unit, positioning a zero equivalence point through a discrete winding number and interpolation, and reconstructing a vortex thread structure after flow field evolution. According to the method, the parallel computing advantage of quantum computing in high-dimensional data processing is fully exerted, the computing efficiency of flow field simulation is remarkably improved, the data processing speed is increased, and meanwhile the high requirement for storage resources is effectively relieved.
Owner:PEKING UNIV +2

Multi-modal practical dynamic security domain construction method and system based on maximum Lyapunov exponent theory

The invention discloses a multi-modal practical dynamic security domain construction method and system based on the maximum Lyapunov exponent theory, and the method comprises the steps: obtaining the measurement information of the active power, the power angle, the angular velocity and the like of a generator node from an actual power system, and determining a fault line and a fault state corresponding to a required dynamic security domain; transient simulation of an initial operation point of the system is carried out to obtain a transient stability condition and a power angle leading cluster of the system; for generators in the leading cluster S, calculating to obtain a maximum Lyapunov exponent characteristic value and related characteristics of a control variable, and solving the maximum Lyapunov exponent sensitivity; and constructing a multi-modal practical dynamic security domain based on the maximum Lyapunov index by combining the upper limit and the lower limit of generator output and the output limit of a balancing machine. According to the method, accurate transient stability evaluation before a fault occurs is realized, the application range is expanded, and the limitation that the stability margin cannot be quantified by track stability judgment is changed.
Owner:ELECTRIC POWER SCI & RES INST OF STATE GRID TIANJIN ELECTRIC POWER CO +2

A silicon carbide grain size detection method based on image processing

The application discloses a silicon carbide grain size detection method based on image processing and relates to the technical field of semiconductor material quality detection. A first feature map is generated by calculating the phase consistency distribution of a silicon carbide metallographic image to suppress scratch interference. A topological anchor sequence is identified by using the eigenvalues of a Hessian matrix, and a balanced first gradient potential field matrix is generated by combining local entropy density to suppress polycrystalline contrast fluctuation. Under the constraint of the topological anchor, an optimization evolution is performed by using a path cost function with curvature penalty and energy saturation characteristics, physical inertia closure of a grain boundary signal fracture is realized, and a closed grid vector is generated. Twin grain boundaries are identified and logically merged based on geometric parameters, and grain size detection data consistent with the physical structure of the material is output.
Owner:SHENZHEN MULINSHENG MICROELECTRONICS CO LTD

Method and device for rapidly calculating sensitivity of linear periodic time-varying system based on matrix sparse technology

PendingCN120744296AComplex mathematical operationsAlgorithmSystem matrix
The invention discloses a rapid calculation method and device for the sensitivity of a linear periodic time-varying system based on a matrix sparse technology, and belongs to the technical field of modeling analysis of the linear periodic time-varying system.The rapid calculation method comprises the steps that an LTP system is converted into an equivalent LTI system, and the eigenvalue, the left eigenvector and the right eigenvector of the LTP system are obtained; based on a functional derivative method, obtaining a sensitivity analytical expression of the LTP system characteristic value to the parameter; according to the influence mechanism of the parameters on the system matrix, obtaining an analysis form of the influence of the parameters on the LTP system matrix in the sensitivity analysis expression; and converting a partial derivative of a system matrix to a steady-state trajectory into sparse calculation of a vector field Hessian matrix by using a matrix sparse technology, thereby obtaining a calculation result of the characteristic value sensitivity of the LTP system. According to the method, the sparse characteristic of the system matrix in the PEPS is utilized, the calculated amount and the storage requirement of the system matrix on trajectory partial derivative solving are reduced, and therefore calculation of the LTP characteristic value sensitivity is accelerated.
Owner:HUAZHONG UNIV OF SCI & TECH

Circuit fault diagnosis method based on fractional order observer

PendingCN121348037AElectronic circuit testingCapacitanceHat matrix
The invention discloses a circuit fault diagnosis method based on a fractional order observer, and the method comprises the steps: constructing a fractional order circuit model based on the Kirchhoff's law, and describing a capacitor end voltage and an inductive current as a Caputo fractional order derivative form; a double-disturbance decoupling mechanism is designed, total disturbance is decomposed into a decoupling disturbance component and a non-decoupling disturbance component, physical isolation of the decoupling disturbance is achieved through a projection matrix, and the influence of the non-decoupling disturbance is eliminated through a dynamic suppression coefficient. Constructing a state observer to generate a state estimation error and a fault error vector, and dynamically compensating a fault estimation value in combination with an adaptive gain matrix; and ensuring the system stability through a Lyapunov function, and proving that an error state is converged to a preset boundary domain under the condition of satisfying a characteristic value correlation matrix inequality. Compared with the prior art, rapid and accurate diagnosis of faults such as inductance saturation and capacitance aging can be realized.
Owner:HUAIYIN INSTITUTE OF TECHNOLOGY

Intervention device, intervention method, and program

To provide an intervention device capable of estimating which factors governing the state of the system should be intervened in order to return the system to then original stable state just before the system transitions from the stable state to a critical state.SOLUTION: An intervention device 1 for suppressing fluctuations in networks where nodes with state quantities are connected includes: a state sample acquiring unit 11 that acquires multiple samples of state; an estimation unit 12; an intervention unit 13; and a control unit 14. The estimation unit 12 selects eigenvectors related to the eigenvalues of the covariance matrix of the acquired multiple samples as an estimate of the dominant right eigenvector. The control unit 14 is configured to control the intervention unit 13 so as to intervene to suppress the state quantity corresponding to the eigenvector component related to the selected eigenvalue on the network.SELECTED DRAWING: Figure 5
Owner:THE JAPAN SCI & TECH AGENCY +1

System and Method for Spectral Learning in Cognitive Manifolds

PendingUS20260187531A1AlgorithmZ eigenvalue
A system and method for spectral learning in persistent cognitive machines implements learning through controlled evolution of a spectral decomposition of a cognitive manifold. The spectral decomposition, comprising eigenvectors and eigenvalues, encodes long-term memory as global geometric structure rather than as stored data or network parameters. The system performs inference operations by projecting incoming data onto the cognitive manifold using a fixed spectral decomposition without modification. Geometric invariants including principal angles, spectral gap ratios, projection residuals, and curvature statistics are continuously monitored to detect structural inadequacy. When invariants exceed thresholds, a learning event modifies the spectral decomposition through eigen decomposition with warm-start initialization while enforcing mode-specific plasticity bounds that are tighter for low-frequency eigenvectors than high-frequency eigenvectors, thereby preventing catastrophic forgetting. The system operates continuously by alternating between inference using fixed spectral decompositions and learning events that modify spectral decompositions through controlled spectral evolution.
Owner:ATOMBEAM TECH INC

A graph data semantic analysis method, device, equipment and readable storage medium

The application discloses a kind of graph data semantic analysis methods, this method will graph data semantic analysis process be disassembled into three parts of calculating N order inner eigenvalue, N order outer neighbor feature and feature combination, the calculation of neighbor feature most consumed resource is moved to offline stage, when receiving the data analysis request initiated by user, when certain node is as the N order neighbor node of target node, take out the X order historical neighbor feature pre-cached, the whole flow of deep neighbor modeling calculation can be recovered, can be guaranteed to analyze the depth as N+X order neighbor feature, while, the query order of online graph database is reduced to N order neighbor from N+X order neighbor, greatly reduce the query pressure of online graph database, guarantee the calculation of online stage, through the light degree of data to speed up the calculation speed, reduce the consumption of computing resources.The application also discloses a kind of graph data semantic analysis device, equipment and readable storage medium, with corresponding technical effects.
Owner:DUXIAOMAN TECH (BEIJING) CO LTD

A method and device for improving numerical shock stability of hypersonic speed

The application relates to a kind of computing method and device for improving the numerical shock stability of hypersonic speed, belonging to the technical field of computational fluid dynamics.The method comprises: obtaining the grid information of numerical simulation of hypersonic vehicle, and initializing flow field information; for each calculation grid unit, two calculation localities are constructed along the x direction and the y direction with two adjacent grids; two localities are respectively equipped with stability matrix, and the instability condition of grid in the calculation process is judged by the positive and negative of the eigenvalue thereof; according to the instability condition of the grid unit, a Riemann solver is selected: for the grid unit not prone to instability, a high-resolution HLLC Riemann solver is used for calculation; for the grid unit prone to shock instability, a high-robustness HLL Riemann solver is used for calculation. By using different Riemann solvers for calculation at different positions, the shock instability position in the flow field can be detected, identified and processed, and the stable capture of shock is realized.
Owner:NAT UNIV OF DEFENSE TECH

New energy unit clustering method based on modified node admittance matrix

ActiveCN117251746BGeometric CADDesign optimisation/simulationLaplacian spectrumAlgorithm
The application provides a new energy unit clustering method based on a modified node admittance matrix, comprising the following steps: (1) reading power grid topology data and element parameters to obtain a branch table of the power grid; (2) constructing a modified node admittance matrix according to the branch table of the power grid; (3) calculating eigenvalues and eigenvectors of the modified node admittance matrix; (4) sorting n eigenvalues in ascending order to obtain new eigenvalues and corresponding eigenvector sequences; and (5) based on the new eigenvalues and corresponding eigenvector sequences obtained in step (4), using a mean value division method based on Laplacian spectrum to divide nodes in the power grid. The network topology is described by the constructed modified node admittance matrix, and the nodes are divided by the mean value division method based on the Laplacian spectrum, so that the new energy units can be accurately clustered. The application is suitable for grouping of various new energy stations.
Owner:HUBEI XINNENG ZHICARBON ENG TECH CO LTD

A method for analyzing dynamic mode and flow stability of double-box-girder flow field

ActiveCN120874179BAccurate numerical simulationSimulation results are reliableGeometric CADSustainable transportationBridge engineeringComputational model
The application discloses a kind of double box girder flow field dynamics modal and flow stability analysis method and system, belong to bridge engineering field.This method first establishes double box girder flow field model by computational fluid dynamics, solves the momentum equation and continuity equation of two-dimensional incompressible flow;Dynamic modal decomposition is carried out again, and eigenvalue, characteristic mode and other information are obtained;Finally, based on linear stability theory, the flow field component is expressed as the sum of steady-state solution and disturbance term, and the global linear flow stability equation is obtained by substituting into the equation, and the flow stability is analyzed by discrete.This method can accurately extract the modal characteristics of each order, providing a scientific basis for wind-induced vibration analysis of double box girder, and has important application value.The method of the application occupies less computing resources, has high numerical accuracy, and can be further expanded to higher precision calculation model according to computing resources, and is suitable for application in actual bridge engineering.
Owner:HARBIN INST OF TECH

Road damage detection method based on vision assistance

The invention relates to the technical field of computer vision, and discloses a road damage detection method based on vision assistance. The method comprises the following steps: firstly, collecting a pavement grayscale image, and establishing a pixel coordinate and a row-column index with the left upper corner as an original point; gaussian smoothing is carried out on the image, and first-order partial derivative is solved in row and column directions to obtain a gradient field; carrying out weighted accumulation on the gradient in a local window to form a symmetric structure tensor, and taking a characteristic value difference as structure energy; second-order and mixed partial derivatives are solved through second-order difference, and curvature tensor is constructed to obtain curvature energy; standardizing the two types of energy, and weighting the two types of energy into coupling energy according to a self-adaptive proportion; setting a threshold value according to the total graph statistics to generate candidate masks; and outputting structured results such as boundaries, geometric centers and the like through four neighborhood connected domain and area and shape ratio screening. The method gives consideration to cracks and pits, improves the accuracy and adaptability, reduces the manual dependence, and is convenient for large-scale batch application.
Owner:NUCLEAR IND JINHUA ENG EXPLORATION INSTITUTIONS

Seismic data adaptive denoising method and system based on quantum mechanics

The invention discloses a seismic data adaptive denoising method and system based on quantum mechanics, and belongs to the field of seismic data denoising, and the method comprises the following steps: mapping seismic data into a quantum potential energy field based on the predictability of a seismic signal in a frequency-space domain, a self-adaptive quantum basis function is constructed by solving a characteristic value problem of a Schrodinger equation; performing sparse representation on the seismic data by adopting a quantum basis function to obtain an expansion coefficient of the signal on a quantum basis; performing noise suppression on the expansion coefficient by adopting an energy-based threshold processing mechanism to obtain a denoised coefficient; and reconstructing de-noised seismic data through quantum basis function linear combination by using the de-noised coefficient. The method can be seamlessly connected with an existing seismic data processing flow, is convenient for seismic data processing personnel to use, and improves the analysis precision of seismic data.
Owner:SANYA MARINE OIL & GAS RESEARCH INSTITUTE NORTHEAST PETROLEUM UNIVERSITY

Large-scale power system discrete eigenvalue parallel computing method based on surrounding channel integration

The invention relates to a large-scale power system discrete eigenvalue parallel computing method based on surrounding channel integration. Comprising the following steps: performing contour integral spectrum transformation on a system state matrix, and determining characteristic value dominance; differentiated integral curves are designed based on characteristic value dominance, the number of characteristic values in each integral curve is estimated, and the initial dimension of the corresponding characteristic subspace is determined; determining the number of initial integral points of each integral curve by adopting a self-adaptive method, and performing self-adaptive integral point configuration on each integral curve; determining an initial subspace dimension of each block based on the number of feature values and the number of parallel integral blocks; performing singular value decomposition on a block basis matrix to construct a standard orthogonal basis; combining the orthogonal basis of each block to form a global orthogonal subspace basis; performing Rayleigh-Ritz projection in a global space to solve a dimension-reduced generalized feature value; and outputting the discrete eigenvalue and the eigenvector. And accurate and efficient calculation of discrete characteristic values of a large-scale power system is realized.
Owner:SICHUAN UNIV +1

Vector inner product operation method and device and electronic equipment

The invention discloses a vector inner product operation method and device and electronic equipment, and belongs to the technical field of computers. The method comprises the following steps: acquiring an input first vector; comparing each first feature value in the first vector with a pre-generated random number sequence to obtain a first bit stream corresponding to each first feature value; obtaining a first bit stream corresponding to each second bit stream according to the position of a second feature value corresponding to a plurality of pre-stored second bit streams in a second vector, and performing bitwise AND operation on the obtained first bit stream and the corresponding second bit stream to obtain a third bit stream corresponding to each second feature value; and determining an inner product operation result of the first vector and the second vector according to the number of preset bit values in the third bit stream corresponding to each second feature value. In this way, the probability of bit flipping can be reduced, and the accuracy of vector inner product operation can be improved.
Owner:ZTE CORP

A method, device, electronic device and storage medium for graph data partitioning based on label characteristics

The present application discloses a graph data partitioning method, device, electronic device and storage medium based on label characteristics, the method comprising: extracting N weakly connected components from the graph data; wherein a weakly connected component comprises a descriptive label between entity objects in the graph data, and N is an integer greater than 1; calculating the clustering coefficient and the minimum non-zero eigenvalue of the first weakly connected component of any one of the N weakly connected components in M ​​preset partitions; wherein the clustering coefficient is used to characterize the label density of the first weakly connected component in each of the M preset partitions, and the minimum non-zero eigenvalue is used to characterize the connectivity of the first weakly connected component in each of the M preset partitions; wherein M is an integer greater than 0; and determining the target partition of the first weakly connected component in the M preset partitions based on the clustering coefficient and the minimum non-zero eigenvalue of each of the M preset partitions.
Owner:NAT UNIV OF DEFENSE TECH

Implementable k-entanglement measurement and numerical calculation method independent of convex top continuation

The invention relates to the technical field of quantum information and data processing, in particular to a practical k-entanglement measurement method and numerical calculation method independent of convex top continuation. The entanglement metric calculation method is based on a k-entanglement witness method and a power iterative numerical algorithm, and the core of the method is to estimate the maximum eigenvalue of a matrix and the eigenvector corresponding to the maximum eigenvalue through an iterative numerical method. According to the method, the maximum separable feature value and the corresponding separable feature vector are solved through power iteration, and then the calculation process is further optimized in combination with forward iteration and backward iteration, so that the accuracy and stability of the result are ensured. The whole solving process is intelligent and efficient, and a database of the maximum separable characteristic value and the corresponding quantum state can be established. The database is a basis for constructing a corresponding k-entanglement witness. According to the method, the calculation process is greatly optimized, the feasibility and reliability of numerical calculation are remarkably improved, and a practical, efficient and accurate entanglement measurement tool development method is provided for users.
Owner:TAIYUAN UNIVERSITY OF TECHNOLOGY

A method and system for identifying dumpling micro-cracks

PendingCN122473149AAlgebraic connectivityDecision model
The present application provides a kind of dumpling micro crack identification method, system, the identification method includes collecting the image of the region to be measured, fusion pixel main curvature tensor and multi-direction phase consistency feature, constructs multi-scale feature pyramid;Rely on the multi-scale micro crack membership degree feature map generated by convolution network containing channel attention, weighted fusion obtains main membership degree atlas, calculates threshold screening pixel average membership degree as main membership value, screening pixel constructs binary connected domain and skeleton, extracts topological node, constructs local topological atlas according to the difference of node quantity;Solve Laplace matrix eigenvalue to obtain algebraic connectivity, combined with the eigenvalue proportion of covariance matrix to calculate linear structure degree, fusion three core indexes input classification decision model, realize accurate identification of dumpling surface micro crack.
Owner:PUYANG ZENGYUN FOOD CO LTD

A method and system for detecting vulnerability of a multi-sensor network topology

This application discloses a method and system for detecting the vulnerability of multi-sensor network topology, relating to the field of sensor technology. The method includes: calculating eigenvalues ​​of the state matrix. l i ( A Construct a weighted directed topological graph, establish the Laplace matrix, and obtain eigenvalues ​​through eigenvalue decomposition. m j ( L ); Calculate the state estimates of each sensor and stack them to obtain the overall system error; Establish the state matrix Φ of the unified consensus estimation error system and find the joint eigenvalues ​​of Φ; Construct the deviation system to form the reachability matrix. R If eigenvalues ​​exist m j ( L ) and eigenvalues l i ( A This allows for the simultaneous joint eigenvectors. v ij Belongs to reachable subspace span ( R If the current multi-sensor network topology and consensus gain configuration are vulnerable to Byzantine attacks, then the present application's method can determine the vulnerability of the topology without extensive time-domain simulations or physical attack experiments.
Owner:YANAN UNIV

Method and system for stability determination of linear periodic time-varying system based on harmonic state-space model, and medium

This invention discloses a method, system, and medium for determining the stability of a linear periodic time-varying system based on a harmonic state-space model, belonging to the field of linear system dynamics analysis. The method includes: constructing harmonic state-space models M with high truncation order H and low truncation order L for the system, respectively. H M L ; Calculate M L The first eigenvalue and the first eigenvector; for M H Arrange its state variables in the order of to transform them; calculate M under the first eigenvalue. L The system matrix relative to the transformed M H The deviation of the system matrix is ​​used to obtain the deviation matrix; the eigenvalue deviations are calculated based on the deviation matrix and the first eigenvector; the sum of the eigenvalue deviations and the first eigenvalue is calculated to obtain M. H The second eigenvalue is used to determine the stability of the linear periodic time-varying system. This approach ensures the accuracy of eigenvalue calculation while improving computational efficiency, thus enhancing the applicability of the harmonic state-space model to large-scale systems.
Owner:HUAZHONG UNIV OF SCI & TECH

Data principal acquisition method based on transverse federated learning

The present application relates to the field of information technology, and more particularly to a data principal component acquisition method based on transverse federated learning, comprising: respectively obtaining features and vectors of local sample data; participating parties negotiate to generate a random number vector; adding the features and the vectors and sending to a trusted coordinator; the trusted coordinator calculates the mean; the participating parties calculate the difference and the covariance matrix; again generating a random number vector, adding the covariance matrix and sending to the trusted coordinator; calculating the global covariance matrix; obtaining m eigenvalues and eigenvectors of the covariance matrix; selecting d eigenvalues and corresponding eigenvectors from the m eigenvalues in descending order and sending to each participating party; the participating party projects the local sample data into a d-dimensional space formed by the eigenvalues to obtain a projection, which is the principal component of the local sample data. The beneficial technical effects of the present application include: while protecting data privacy, allowing more sources of data to be used, and improving the accuracy of model analysis.
Owner:HUZHOU XINYUN TECH CO LTD

New energy unit grouping method based on admittance laplacian matrix and multi-dimensional binary coding quadrant division

The application provides a new energy unit grouping method based on an admittance Laplacian matrix and a multi-dimensional binary coding quadrant division, and comprises the following steps: reading power grid topology data and element parameters to obtain a branch table of the power grid; constructing an admittance Laplacian matrix according to the branch table of the power grid; calculating eigenvalues and eigenvectors of the admittance Laplacian matrix; sorting n eigenvalues in ascending order to obtain new eigenvalues and a corresponding eigenvector sequence; and dividing nodes in the power grid by using a Laplacian spectrum multi-dimensional binary coding quadrant division method. The network topology is described by the constructed Laplacian matrix, the nodes in the power grid are divided by using the Laplacian spectrum multi-dimensional binary coding quadrant division method, and the new energy unit can be accurately grouped. The application is suitable for grouping and clustering of various new energy stations.
Owner:STATE GRID HUBEI ELECTRIC POWER RES INST +2

A coherent signal DOA estimation method based on coprime array

The present invention discloses a coherent signal DOA estimation method based on a coprime array. First, the covariance matrix of the coprime array is calculated, and eigendecomposition is performed on it to find the eigenvector corresponding to the maximum eigenvalue. The eigenvector is then decomposed into two parts according to the number of array elements of the coprime array, and the elements of the two parts are rearranged to obtain two Hankel matrices H1 and H2. Next, singular value decomposition is performed on H1 and H2 to obtain the noise subspaces of H1 and H2. Finally, based on the noise subspaces of H1 and H2, the spatial spectra of H1 and H2 are plotted using the MUSIC method. The same spectral peaks in the spatial spectra of H1 and H2 are found, and the DOA values ​​corresponding to the spectral peaks in the spatial spectra of H1 and H2 are recorded and averaged to obtain the final DOA estimate. The method of the present invention does not require vectorization of the received signal of the sparse array, can fully utilize the coprime characteristics of the array, and obtain the angle estimate of the coherent signal, thus filling the gap in coherent signal arrival angle estimation for coprime arrays.
Owner:NANJING UNIV OF AERONAUTICS & ASTRONAUTICS

A pruning optimization method for large language models

The application discloses a pruning optimization method for a large language model, comprising: an inter-layer feature spectrum acquisition step, based on PCA or eigenvalue decomposition, estimating the amount of information that can be preserved in each layer under different reserved dimensions; a gradient sensitivity acquisition and logarithmic mapping calibration step, using gradient to describe the influence of parameter change, and weakening resource allocation imbalance through logarithmic mapping; and a global non-uniform sparsity allocation step, allocating the reserved dimensions according to the principle of maximum marginal benefit. The application solves the problem that the traditional pruning method ignores the inter-layer heterogeneity and the long-tail distribution of gradients, leading to resource allocation imbalance, realizes a sparse configuration that is more in line with the inter-layer heterogeneity of the model, is suitable for edge computing devices and low-memory server deployment, and effectively reduces resource consumption under the premise of ensuring the performance of the model.
Owner:SHANGHAI UNIV

LTP system feature vector calculation method for power system stability analysis and application

The invention discloses an LTP system feature vector calculation method for power system stability analysis and application, and belongs to the field of linear periodic time-varying system analysis in a power system.The LTP system feature vector calculation method comprises the steps that dynamic differential equations of a left feature vector and a right feature vector of LTP system feature values are obtained; establishing a relation among the LTP system characteristic value, the left characteristic vector, the right characteristic vector and the state transition matrix, and deducing to obtain solutions of the left characteristic vector and the right characteristic vector at 0 and T moments; based on the differential equation of the right feature vector, deriving a solution of the right feature vector in a period T through a forward modeling form; and based on the state transition matrix and the differential equation of the left eigenvector, deriving in a time inversion form to obtain a solution of the left eigenvector of the eigenvalue in the LTP system in a period T. Through the LTP system feature vector calculation method disclosed by the invention, the requirement for ill-conditioned STM inversion is avoided, and more reliable left and right feature vectors are obtained.
Owner:HUAZHONG UNIV OF SCI & TECH

Mechanical arm heterogeneous task cooperative control method based on graph neural network

The invention discloses a mechanical arm heterogeneous task cooperative control method based on a graph neural network, and the method comprises the following steps: obtaining a plurality of heterogeneous tasks, and constructing a task relation graph; carrying out feature decomposition to obtain a Laplacian feature value set and a Laplacian feature vector set; coding is carried out, graph structure propagation processing is executed, and spatial domain task embedding representation is generated; projecting the spatial domain task embedded representation to a spectral domain to generate a spectral domain task representation; executing improved neural Laplacian operator processing, and performing weighting and adjustment to obtain spectral domain response representation; performing frequency domain energy modulation to obtain spectral domain control representation; inverse spectral domain transformation is executed, and time domain cooperative control representation is generated; the time domain cooperative control is converted and expressed as a mechanical arm action control quantity, and a control instruction is output; and executing feedback to construct a frequency domain error, and executing adaptive updating. According to the invention, the graph neural network and spectral domain regulation are adopted to realize the cooperative control of the heterogeneous task of the mechanical arm, and the control stability is high.
Owner:SHANGHAI ZHIYISHEN ROBOT CO LTD

High-precision position keeping method and device for multi-dimensional motor cooperative control and storage medium

The invention discloses a high-precision position keeping method and device for cooperative control of a multi-dimensional motor and a storage medium. The high-precision position keeping method and device are used for keeping the high-precision position of the multi-dimensional motor. The method comprises the following steps: constructing a multi-dimensional data set; performing standardization processing on the multi-dimensional data set to obtain standard data; calculating a covariance matrix based on the standard data and performing eigenvalue decomposition; determining a plurality of principal component feature vectors according to the cumulative variance contribution rate to form a principal component space; projecting the standard data into the principal component space to form a dimensionality reduction data set; inputting the dimension reduction data set into a pre-trained deep belief network, and outputting a comprehensive feature vector through hierarchical features of a restricted Boltzmann machine; inputting the comprehensive feature vector into a depth deterministic strategy gradient algorithm model, and constructing a reward function; outputting a real-time adjustment strategy based on the reward function, and obtaining an optimization control parameter; and realizing high-precision position keeping of the plurality of motors according to the optimized control parameters.
Owner:雷文斯(深圳)科技有限公司