41 results about "Eigenvalue distribution" patented technology

The invention provides a signal blind separation and intelligent reconstruction method and system in a complex scene, and relates to the technical field of signal processing, and the method comprises the steps: receiving an aliasing signal, and converting the aliasing signal into a multi-channel signal observation matrix; performing decomposition in a wavelet domain to obtain a wavelet coefficient, matching the wavelet coefficient with the sparse dictionary, and reconstructing a target signal source after optimization; estimating the number of signal sources based on covariance matrix eigenvalue distribution; mapping a signal source to a graph structure domain, extracting space and time sequence correlation through a mixed graph convolutional network, and obtaining a target separation signal through variational reasoning optimization; and finally, carrying out quality evaluation and post-processing to obtain a final reconstruction signal. According to the invention, the signal separation precision and robustness in a complex scene are improved.

The invention provides a theoretical modeling method and system for describing the relationship between the gradient entropy and the compression rank, and provides a theoretical basis for gradient compression strategy design and adaptive communication optimization. According to the method, modeling is carried out on gradient matrix eigenvalue distribution, internal relation between eigenvalues and compression errors is revealed by applying a random matrix theory and characteristic spectrum analysis, and a function relation between a compression rank and a reconstruction error is constructed. On the basis, a constraint condition that a compression error absolute value is kept constant is introduced, the standard deviation of the gradient matrix is incorporated into error estimation, a corresponding relation between the standard deviation and a compression rank is derived, and the effect of the standard deviation serving as a compression error intermediate variable is clarified from the statistical angle. According to the method, the minimum compressible rank under specific precision can be estimated, a unified analysis framework is provided for analyzing compressibility differences of different training stages or network layer gradients, and the method has good theoretical universality and engineering practicability.

The invention provides a multi-sensor joint optimization SLAM method and system based on adaptive weight. The method comprises the following steps: synchronously acquiring observation data of a laser radar, a visual sensor and an inertial measurement unit; determining the degradation state of each sensor, and correcting the uncertainty matrix of state estimation based on the degradation determination result; calculating information entropy based on the corrected eigenvalue distribution of the uncertainty matrix of each sensor, and combining determinant calculation of the matrix to obtain a quantitative reliability evaluation value; dynamically generating a fusion weight according to the quantitative reliability evaluation value so as to suppress the influence of the sensor which is judged to be degraded; and carrying out weighted fusion on the observation residual errors of the multiple sensors based on the fusion weight, and solving the optimal pose and map estimation of the system.

The invention discloses an enterprise credit evaluation analysis method, and relates to the field of data processing, and the method comprises the steps: obtaining the credit data of a plurality of to-be-evaluated enterprises; performing feature extraction on the credit investigation data through an improved stack type self-encoding neural network model to obtain feature data; the improved stack type self-encoding neural network model comprises an encoder, a decoder, an attention layer and a prior rule layer; the attention layer adopts a Scaled Dot-Product Attention mechanism to learn a weight feature matrix Ha, and the priori rule layer sets feature constraints according to a priori rule; clustering the feature data to obtain a cluster to which the enterprise belongs; according to the feature value distribution of the enterprises in the clustering clusters, performing feature scoring by using a prior rule to obtain clustering labels of the clustering clusters; according to the clustering label and the weight feature matrix Ha, generating an enterprise credit investigation portrait, and according to the enterprise credit investigation portrait, carrying out credit investigation evaluation analysis; aiming at the low enterprise credit evaluation precision in the prior art, the reliability of the evaluation result is improved.

The invention discloses a cable surface defect monitoring method based on image recognition, and relates to the technical field of industrial nondestructive testing, and the method comprises the steps: carrying out the two-dimensional Fourier transform of a cable surface visible light image, extracting a spatial frequency corresponding to a spectrum main peak as a lattice basis vector, and building a space-time crystal model; performing spatial frequency mapping in a reciprocal space of the space-time crystal model, and positioning boundary points of a Brillouin region; analyzing the characteristic value distribution of the group representation matrix at the boundary points of the Brillouin region, identifying the space coordinate position of the symmetry defect, and marking the space coordinate position as a potential defect region; and based on the equivalent divergence distribution diagram, the three-dimensional defect form and depth distribution, executing space trajectory tracking on the identified symmetric defect, screening out isolated abnormal points caused by transient interference, and outputting a monitoring report. According to the method, the mapping relation between the divergence value and the defect depth in the equivalent divergence distribution diagram is established through the Poisson equation, and inversion of the three-dimensional shape and depth distribution of the potential defect area is achieved.

The invention discloses a small signal stability analysis method and system for a large-scale new energy power system, and relates to the technical field of power system stability analysis, and the method comprises the steps: constructing a discrete state space model of the whole power system for the large-scale new energy power system, and determining a discrete state matrix; converting a characteristic value of a system state matrix expected to be solved into a system dominant characteristic value by utilizing a preprocessing technology; the preprocessing technology is displacement inverse transformation or rotation transformation; based on the discrete state matrix, calculating and solving a system dominant characteristic value by using an iterative algorithm in combination with sparse implementation; converting the solved dominant characteristic value of the system into a characteristic value of a continuous state matrix, judging the small signal interference stability of the power system according to the converted characteristic value, and comparing the converted characteristic value with a distribution diagram of all the characteristic values of the system for verification. According to the method, the problems of difficulty in solving discrete space state matrix eigenvalues and low calculation efficiency are solved.

The invention relates to a storage calculation device supporting sparse matrix calculation, which comprises a storage core particle, a logic core particle, a multiply-accumulate circuit, a characteristic value distribution circuit and a sparse matrix decomposition circuit, and is characterized in that the storage core particle is used for storing a sparse weight matrix, and the logic core particle is used for data interaction, caching and control; the multiply-accumulate circuit executes multiply-accumulate operation of characteristic values and weights, the sparse matrix decomposition circuit screens out position codes of front # MAC non-zero weights in each operation period and outputs control signals, and the characteristic value distribution circuit dynamically maps input characteristic values to corresponding non-zero weights according to the control signals, so that calculation of an arbitrarily distributed sparse matrix is realized. The method has the advantages that multiply-accumulate operation is performed on the non-zero weight, the zero weight is skipped, power consumption is reduced, calculation efficiency is improved, and efficient processing of the sparse matrix is achieved through a storage and calculation integrated framework.

The present invention discloses a process reliability evaluation method for a spiral bevel gear milling machine tool, comprising the following steps: Step 1: establishing an instantaneous cutting force model and a milling dynamics model for processing spiral bevel gears based on a forming method, and calculating the cutting vibration characteristics of the spiral bevel gear milling machine tool during the milling process; Step 2: based on the cutting vibration characteristics in Step 1, evaluating the sensitivity of various influencing factor parameters to the cutting vibration; and calculating the amplitude eigenvalue distribution type; Step 3: based on the amplitude eigenvalue distribution type in Step 2, using the Bayesian formula, establishing a BP neural network model, and using the Garson algorithm to update the BP neural network model; using the updated BP neural network model to calculate the amplitude eigenvalue under extreme working conditions, and combining the vibration trajectory curve of the milling tool and the process reliability formula to complete the machine tool process reliability evaluation. The present invention also discloses an optimization method for tuning machine tool process parameters using the above evaluation method.

The invention discloses a DC power system fault detection method, system and device, and a storage medium, and relates to the technical field of fault detection. The method comprises the following steps: collecting node data in a direct-current power system, and generating a column vector according to a time sequence; performing standardization processing on the column vector to obtain a standard non-Hermitian matrix; calculating a standard matrix product and a corresponding characteristic value of the standard non-Hermitian matrix, and determining characteristic value distribution of the standard matrix product according to a single-ring theorem; the characteristic value distribution is limited by the inner ring radius and the outer ring radius; constructing a random variable based on the characteristic value distribution and the average spectral radius; when the random variable is smaller than the radius of the inner ring, judging that the system breaks down; and when the random variable is greater than or equal to the radius of the inner ring, determining that the system has no fault. According to the method, the defects of the traditional method in quickness, anti-interference performance and multi-end system adaptability can be overcome by quantifying the linear change trend of the fault correlation characteristics.

The application relates to the technical field of dynamic modeling, in particular to a physical simulation model generation method and system for unstructured environments, which specifically comprises the following steps: converting a CAD model of a physical prototype into a point cloud and voxelizing; constructing first characteristic values of each voxel according to local geometric features of the point cloud in each voxel and overall geometric features of the voxel; adjusting initial grid sizes of each voxel based on differences in distribution positions of the first characteristic values of each voxel in the overall and local neighborhoods, combining the first characteristic values, determining adaptive grid sizes of each voxel, and obtaining a final voxel division result of the CAD model; and performing physical simulation by using the final CAD model voxel division mode; the method solves the problem that a traditional physical simulation model generation method cannot distinguish local areas with obvious unstructured feature differences, thereby reducing the modeling effect; while ensuring the calculation accuracy of key areas, the overall calculation cost is significantly reduced.

The invention provides a system and a method for detecting abnormal characteristics of a rotary machine, and the system comprises a sensing module which is used for obtaining the vibration data of an accessory of the rotary machine; the data processing module is used for processing the vibration data to generate a plurality of vibration time sequence data; the conversion module is used for converting the plurality of pieces of vibration time sequence data into a plurality of pieces of conversion data; and the abnormal feature detection module is used for comparing feature value distributions in the plurality of pieces of conversion data so as to define the accessory corresponding to the conversion data of which the feature value distributions are changed to be abnormal.

This invention proposes a direction-of-arrival (DOA) estimation method based on a sparse circular array using the golden ratio. The method defines the antenna position and three-dimensional coordinates using the golden ratio, constructing a sparse array. A DOA vector is established based on the azimuth and elevation angles, and the path difference and phase difference are calculated. These are then combined with multi-shot data to form a covariance matrix. To improve robustness under low snapshot and low signal-to-noise ratio conditions, adaptive diagonal loading is employed to adjust the noise eigenvalue distribution and enhance the distinguishability between the signal and noise subspaces. Addressing the issues of traditional MUSIC algorithms' accuracy depending on step size and poor real-time performance, a three-step search is designed: a coarse search to locate local maxima, a ternary trigonometric iteration to narrow the interval, and a fine search followed by quadratic surface fitting to eliminate grid errors. This method effectively suppresses sparse array grating lobes, balancing high accuracy and real-time performance.

The present application belongs to the technical field of image recognition, and particularly relates to a diamond quality detection method and system based on image recognition, comprising the following steps: obtaining a diamond image to be detected, calculating the structure tensor in the local neighborhood of each pixel of the diamond image, constructing an initial structure guide map according to the eigenvalue distribution, and recording the difference between the maximum eigenvalue and the minimum eigenvalue of the structure tensor as an anisotropy degree map; performing multi-scale shear wave transformation on the diamond image, extracting high-frequency subband coefficients at each scale to calculate local energy, and fusing the local energy at each scale to obtain a multi-scale edge saliency map; and fusing the initial structure guide map and the multi-scale edge saliency map to generate a composite guide map. The present application avoids the edge blurring phenomenon caused by traditional filtering, and improves the completeness of diamond surface defect extraction and the accuracy of overall quality detection.

The present application is to solve the problems of spectrum leakage, large loss of useful signal and insufficient interference locking precision in the existing narrow-band interference suppression technology; a narrow-band interference suppression method for Beidou navigation is provided, comprising the following steps: S101: discrete signal sequence sampling and segmented caching; S102: constructing an adaptive signal covariance matrix; S103: performing fast eigenvalue decomposition on the covariance matrix; S104: determining the number of interference components based on the eigenvalue distribution; S105: constructing an interference subspace projection matrix; S106: performing orthogonal projection to suppress narrow-band interference; S107: signal gain compensation and output reconstruction; S108: detecting whether the interference environment has changed dramatically; all components pointing to the interference in the signal space are filtered through the orthogonal projection matrix, avoiding the convergence problem of time-domain adaptive filtering, and effectively dealing with the strong narrow-band interference scene with high power density.

The present invention discloses a bearing fault detection method based on the Gershgorin circle theorem, which relates to the technical field of fault diagnosis. The technical solution identifies abnormal eigenvalues ​​T at an early stage and issues early warnings by establishing an autocorrelation matrix and calculating the eigenvalue T distribution. Secondly, the technical solution overcomes the dependence of traditional methods on high sampling rates, and uses fast Fourier transform and noise filtering technology to simplify data processing and improve system efficiency. Finally, by calculating and evaluating the health index Rhzs, the dependence on experience is reduced, making the detection process more automated and data-driven. The evaluation results are based on historical data and health threshold classification, which improves the accuracy and reliability of fault detection. Overall, the solution optimizes the effect of bearing fault detection through precise data processing and automated analysis.

The application discloses a cable surface defect monitoring method based on image recognition and relates to the technical field of industrial nondestructive testing, which comprises the following steps: performing two-dimensional Fourier transform on a visible light image of a cable surface, extracting a spatial frequency corresponding to a main peak of a frequency spectrum as a lattice base vector, and establishing a space-time crystal model; performing spatial frequency mapping in a reciprocal space of the space-time crystal model, and locating a boundary point of a Brillouin zone; analyzing eigenvalue distribution of a group representation matrix at the boundary point of the Brillouin zone, identifying a spatial coordinate position of a symmetry breaking point, and marking the symmetry breaking point as a potential defect area; based on an equivalent divergence distribution diagram, a three-dimensional defect morphology and a depth distribution, performing spatial trajectory tracking on the identified symmetry breaking point, screening out isolated abnormal points caused by transient interference, and outputting a monitoring report. The application establishes a mapping relationship between a divergence value in the equivalent divergence distribution diagram and a defect depth through a Poisson equation, and realizes inversion of the three-dimensional morphology and the depth distribution of the potential defect area.

The present disclosure provides an approach of computing a plurality of feature attribution vectors from a plurality of samples. The approach determines a plurality of low entropy distribution samples from the plurality of samples based on the plurality of feature attribution vectors, and determines a feature value distribution corresponding to the plurality of low entropy distribution samples. Then, the approach identifies a false positive candidate sample based on the feature value distribution and, in turn, constructs a mitigation rule, based on the false positive candidate sample, to mitigate a future false positive sample.

The invention discloses a time domain stability analysis and improvement method for an LLC resonant converter, and the method comprises the steps: solving and generating a Jacobian matrix of the LLC resonant converter according to an extended description function method; solving all characteristic values of the Jacobian matrix; judging the stability of the closed-loop system of the LLC resonant converter according to a Lyapunov indirect method; if the system is stable, outputting a judgment result; if the system is not stable, PI controller parameters are adjusted, or the characteristic value sensitivity of a certain element beta in the matrix is solved, the size of the element beta in the matrix is adjusted based on the characteristic value sensitivity solving result, then the characteristic value is solved again, and the stability is judged. According to the scheme, the system stability influence trend is reflected through time domain analysis, the effects of stability improvement and system instability prevention are achieved, the influence of the load on the stability of the converter is analyzed through the characteristic value sensitivity, the influence trend of the load change on the stability is estimated according to the characteristic value distribution track, the controller acts in advance, and system instability is avoided.

According to the electric power index analysis-based electric power prediction method and system provided by the invention, during network debugging, the basic electric power state detection network can simulate the feature processing performance of the target contrast network, and the basic electric power state detection network can simulate the feature value distribution condition of the target contrast network; therefore, the target power state detection network generated by the representation vector extraction module in the debugged target power state classification network can accurately extract the power state representation vector, and the target power state detection network does not need redundant training data in the establishment link, thereby reducing the network debugging cost, improving the efficiency, and reducing the network debugging cost. Furthermore, the target power state classification network is obtained through transfer learning debugging, so that the target power state classification network can realize the data analysis capability of a complex large model through an exquisite small model, thereby reducing the hardware environment requirement on deployment equipment, and saving the computing power and deployment cost.

The present disclosure belongs to the technical field of hybrid DC power transmission system stability, and particularly relates to a hybrid DC instability discrimination method and system based on mode resonance, comprising: establishing a full-order state space model of a hybrid DC power transmission system, constructing a system dynamics model under the time scale of a DC control system, linearizing the obtained system dynamics model to obtain a linearized model of the hybrid DC power transmission system; calculating eigenvalues of the linearized model and participation factors of the eigenvalues, screening an alternative resonance mode set according to the obtained eigenvalue distribution, and system parameters strongly related to the alternative resonance mode set; modifying system parameters of the hybrid DC power transmission system, drawing root loci of system eigenvalues, judging whether mode resonance occurs and determining resonance strength of the mode resonance; if mode strong resonance occurs and the resonance strength is greater than a set threshold, it is determined that the system will be unstable, and system parameters at the resonance point are recorded.

1. Name of the design product: Graphical user interface for fault detection and classification of electronic devices. 2. Use of the design product: An electronic device. 3. Design key points of the design product: The graphical user interface on the screen. 4. Picture or photo that best shows the design key points: The front view of Design 1. 5. Design 1 is designated as the basic design. 6. Use of the graphical user interface: This design product is used to view the specific situation of fault detection and classification. By performing a correlation analysis on wafer yield data or wafer test data and device data, information on the reasonable range of eigenvalues (Features) and eigenvalue distribution information can be obtained. Interface interaction process: Change state diagram of Design 1: Based on the front view of Design 1, select the eigenvalue that you want to view specifically in the scatter plot or eigenvalue chart in the front view of Design 1, and jump to the change state diagram of Design 1; The interaction method and use of Design 2 are similar to those of Design 1. Based on the front view of Design 2, select the eigenvalue that you want to view specifically in the scatter plot or eigenvalue chart in the front view of Design 2, and jump to the change state diagram of Design 2.

The present invention discloses a small-signal stability analysis method and system for a large-scale new energy power system, relating to the technical field of power system stability analysis, including: for a large-scale new energy power system, constructing a discrete state space model of the entire power system and determining a discrete state matrix; using a preprocessing technique to convert the eigenvalues of the system state matrix to be solved into the dominant eigenvalues of the system; the preprocessing technique is a displacement inverse transformation or a rotation transformation; based on the discrete state matrix, using an iterative algorithm combined with sparsity implementation to calculate and solve the dominant eigenvalues of the system; converting the solved dominant eigenvalues of the system into the eigenvalues of the continuous state matrix, judging the strength of the small-signal interference stability of the power system according to the converted eigenvalues, and comparing and verifying the converted eigenvalues with the distribution diagram of all the eigenvalues of the system. The present invention solves the problems of difficult solution of the eigenvalues of the discrete space state matrix and low calculation efficiency.

The invention discloses a bearing initial degradation point identification method based on a random matrix theory, and the method comprises the steps: 1, collecting vibration data, forming sampling data vectors, and sorting the vectors according to the sampling time to construct a sampling data matrix; 2, separating window data matrixes from the sampling data matrixes by using a real-time separation window analysis method; 3, performing data processing on the window data matrix, and finally constructing a sample covariance matrix to obtain maximum and minimum eigenvalues; 4, obtaining a characteristic value distribution range, a maximum characteristic value and a minimum characteristic value in a normal operation state according to a random matrix theory; and when the window data matrix exceeding the range is found, the current moment corresponding to the window data matrix is the initial degradation point. According to the method provided by the invention, priori knowledge is not needed, and a required conclusion can be obtained only through data. According to the method, feature extraction of different scales is carried out on the acquired signals, the degradation trend of bearing performance can be highlighted, and the prediction precision of the final life prediction stage is improved.

The present invention relates to the field of data processing technology, and in particular to a method and system for predicting capacity of a virtual power plant. The method obtains a historical capacity factor matrix; obtains eigenvalues ​​of the historical capacity factor matrix; obtains a denoised capacity factor matrix according to the eigenvalue distribution of the historical capacity factor matrix; analyzes the lag relationship between each sequence in the denoised capacity factor matrix and a denoised historical capacity information sequence to obtain a marginal characteristic lag matrix; then analyzes the correlation between each sequence in the marginal characteristic lag matrix and the denoised historical capacity information sequence to obtain a modified marginal characteristic lag matrix; completes the prediction of real-time capacity according to the modified marginal characteristic lag matrix, and makes the predicted capacity result more accurate.

The application discloses a kind of based on closed loop efficiency's heterogeneous directed network key node identification method, system and storage medium.The method is first based on node inherent function index and link synergistic conduction efficiency constructs full-factor adjacency matrix, then obtains network eigenvalue distribution by characteristic spectrum analysis, and based on matrix index constructs system efficiency index, to represent the abundance and quality of network closed loop path;Further utilize the product of left and right eigenvector components corresponding to the principal eigenvalue calculates the participation of each node to system efficiency, and calculates the node disturbance sensitivity under the closed loop efficiency change of node removal disturbance, combined with node closed loop participation, form comprehensive importance index, to realize key node identification and sorting.Compared with prior art, the application can simultaneously fuse node function, link efficiency and closed loop structure characteristics, has higher identification precision and stronger explanation ability to key node in heterogeneous directed network, can be applied to the robustness evaluation, resource allocation and structure optimization of complex synergistic system.

According to one aspect, a method for elaborating a machine learning model for the classification of time series signals is proposed comprising obtaining features of training time series signals associated with different indicated classes, calculating a distinction coefficient between classes for each feature and calculating a distinction coefficient between classes for each combination of features from a distribution of the values of the features for each group of time series signals, ranking the features according to the distinction coefficient of each feature and the distinction coefficient of each combination of features, and training the machine learning model by taking as input for this model at least one feature chosen according to the ranking.

The invention discloses a planar array FDA-MIMO radar main lobe interference suppression method. The method comprises the following steps: estimating target signal power through a Capon technology to dynamically adjust a DL loading factor; based on eigenvalue distribution of a sample covariance matrix, an adaptive subspace dimension selection strategy is introduced to balance robustness and main lobe suppression performance under different mismatch conditions; according to the beam former, the projection subspace and the diagonal loading coefficient can be adjusted according to actual condition changes, the problem of steering vector mismatching caused by distance-angle errors, array element position errors, frequency deviation and coherent local scattering can be solved, main lobe interference is effectively restrained while beam robustness is improved, better SINR performance is obtained, and the beam former has a wide application prospect. And the method has wide application value and popularization prospect.

A functional near-infrared spectral signal enhancement and classification method based on spatiotemporal autocorrelation and attention encoding is proposed to address the limitations of limited sample size and insufficient generalization ability of classification models in functional near-infrared spectral signal data. This method is applicable to the auxiliary diagnosis of neurodevelopmental disorders such as attention deficit hyperactivity disorder (ADHD). The method first preprocesses the raw near-infrared spectral data, extracting oxyhemoglobin signals and calculating spatial autocorrelation parameters, channel-level temporal autocorrelation parameters, and the eigenvalue distribution of the functional connectivity matrix. Then, enhanced data is generated based on the spatiotemporal autocorrelation model. Finally, the raw and enhanced data are input into the STEAFNet deep learning model for accurate classification. This invention generates high-quality enhanced data, expands the sample size, and maintains high fidelity. Combined with deep learning, it improves classification accuracy and generalization ability, providing technical support for clinical applications.

The invention relates to the technical field of wireless transmission, and discloses an anti-interference transmission method for a wireless low-power-consumption transmitter. The method comprises the steps of collecting a real-time electromagnetic interference signal frequency spectrum of an environment where a transmitter is located, and extracting multi-dimensional interference characteristics; constructing an interference feature library based on the features, and dividing interference feature clusters according to the feature distribution density and the intensity change trend; calculating a dynamic fluctuation index and a frequency domain concentration index of each cluster, and obtaining a peak frequency drift index and a spectral bandwidth variation index of each cluster; generating an environmental interference entropy value based on the indexes; constructing a transmission signal sequence covariance matrix and carrying out characteristic decomposition to obtain characteristic value distribution, a concentration ratio, a principal component contribution degree related to a discrete state and a discrete coefficient; and a signal transmission purity index is generated by combining the row vector correlation of the covariance matrix, the environment interference entropy value, the principal component contribution degree and the dispersion coefficient, anti-interference transmission control is executed according to the index, and the transmission stability of the transmitter in the complex electromagnetic environment is improved.

This disclosure relates to a multi-party data processing method, a multi-party joint recommendation method, an apparatus, and a readable medium, and relates to the fields of multi-party secure computation and privacy computation. The method determines the intersection potential of the requester's first sample set and the data party's second sample set, filtered by an initial threshold for feature values, through anonymous privacy intersection calculation, without disclosing plaintext information from any party, thus avoiding privacy leaks. Furthermore, it calculates business indicators using the intersection potential and the number of first samples from the first user sample, and evaluates the filtering effect of the initial threshold by the selected intersection quantity to determine a target threshold. This enables the determination of a feature value distribution that fully meets the requester's business needs without disclosing plaintext information. Finally, the data party divides different fourth sample sets based on the target threshold and feature values, corresponding to different business indicator levels of the requester, thereby allowing the data party to support the requester's differentiated business needs in subsequent multi-party collaborations.