24 results about "Spatial covariance matrix" patented technology

The invention provides an ocean wind field prediction method based on a neural network, and belongs to the technical field of ocean wind field prediction.The method comprises the steps that sparse ocean observation data are collected, a spatial covariance matrix is established, the spatial covariance matrix is converted into a graph structure, and then multi-hop neighborhood feature aggregation is conducted through a graph convolutional network; a tensor decomposition algorithm is combined for modeling high-order feature interaction to generate a gridding wind field, a bidirectional long-short-term memory network encoder is used for extracting space-time invariant features, a multi-layer perceptron predictor is used for directly mapping a future multi-step wind field, and a course learning strategy and a Shenchang differential equation boundary layer are matched for correction. The technical problem that sparse ocean observation data are difficult to accurately reconstruct into a high-resolution gridding wind field is solved.

The invention discloses a signal fusion processing method for a motor imagery training brain-computer interface, and particularly relates to the technical field of neural signal processing. Multi-channel scalp electroencephalogram data of a user are collected in real time, a time-varying space covariance matrix sequence is constructed through Riemannian geometric space mapping analysis, and a phase amplitude coupling index sequence between different brain rhythms is analyzed and calculated through cross-band coupling dynamics; performing feature hierarchy joint coding on the spatial covariance matrix sequence and the phase amplitude coupling index sequence to generate a real-time neural representation tensor; calculating the distribution divergence of the real-time neural representation tensor relative to the static reference feature manifold in the initial training stage so as to quantitatively represent the mismatch degree; and finally, dynamically adjusting a weighting coefficient and a fusion structure of the multi-modal signal fusion device according to the characterization mismatching degree, and outputting a motion control instruction matched with the current brain state of the user in real time. According to the method, the motion intention decoding precision and stability in the motion imagination training process are improved.

The present application relates to the technical field of attention mechanism, and discloses an interrogation psychological sensitive point detection method based on a timing attention mechanism, wherein a microphone signal is processed in a local time window through a short-time Fourier transform to generate a spatial covariance matrix time sequence under a symmetric positive definite manifold, and a relative displacement weighted kernel reflecting inter-frame time lag correlation is extracted from a timing attention model output; then, based on an affine invariant metric, a bisection vector is constructed to calculate a bisection cross-sectional curvature density to quantize the geometric inconsistency of the two; subsequently, the weighted kernel is adjusted by the index gate, the geometric aligned predicted spatial covariance matrix is obtained through cut space summation and exponential mapping, the actual and predicted matrix residuals are calculated by measuring the distance of the affine invariant metric, a single time sequence score is generated by energy proportion weighting, and finally, an adaptive threshold is constructed by using the median absolute deviation of the score and a threshold coefficient to screen sensitive moments that meet the conditions, thereby effectively improving the detection accuracy and scene adaptability.

This invention discloses a method for updating a long-service hydraulic twin model based on multi-fidelity Bayesian methods. It involves collecting non-contact, full-field micro-vibration time-history data of hydraulic structures and extracting high- and low-fidelity data observations. Combining macroscopic physical aging equations, a time-varying degradation prior probability distribution with historical memory is constructed. Based on the high- and low-fidelity data observations, a spatially and mechanistically heterogeneous multi-fidelity twin architecture is built. A spatial covariance matrix of full-field measurement points is introduced, and the Bayesian heterogeneous MCMC method is used to infer parameters of the multi-fidelity twin architecture until convergence. The high-confidence posterior parameters of the structural physical parameters and boundary nonlinear stiffness are mapped to the long-service hydraulic twin model, enabling high-frequency adaptive evolution and synchronous updating of the full life-cycle state of the long-service hydraulic structure digital twin under complex uncertainties.

The invention discloses a fan body space wind field reconstruction method based on double-wedge-mirror spiral scanning, and the method comprises the steps: carrying out the spiral scanning through employing a double-wedge-mirror combination, obtaining a scanning track point cloud, and obtaining the radial distance, azimuth angle and pitch angle of each scanning point in a fan body space relative to a wind measurement radar according to the coordinates of the scanning track point cloud; obtaining a trend model of wind speed distribution of an estimation point according to a wind speed measurement value of a scanning point obtained by a radar, and obtaining a spatial covariance matrix between a radar observation value and an estimation value by using an exponential function model representing the spatial correlation of the wind speed of the scanning point; according to a space covariance matrix between a radar observation value and an estimation value and a trend model of estimation point wind speed distribution, the estimation value of the wind speed at a target point is obtained by solving a universal Kriging equation, and the uncertainty of the estimation value is obtained. According to the invention, rapid three-dimensional wind field reconstruction can be realized based on the radial wind speed data of the double-wedge-lens spiral scanning wind field detection laser radar.

The application provides a kind of arid region ecological quality monitoring method and system based on time series spectral remote sensing, it is related to time series spectral remote sensing monitoring technical field, the application is by synchronous acquisition multispectral and thermal infrared image and eliminates water cloud interference, accurately extracts the physical greenness of effective pixel, physical humidity, chemical salinity, apparent dryness and thermal radiation temperature composition physicochemical parameter set;Construct spatial covariance matrix and perform feature decomposition, realize the objective dimension reduction and distribution of parameter weight;According to the positive and negative direction weighted execution of ecological succession indication direction, and using the first eigenvalue carries out spatial baseline flattening, generates regional comprehensive ecological quality evaluation index;Based on the index, construct time series to calculate the time series succession rate, and compare with the degradation classification threshold, realize the objective dynamic determination of arid region overall land surface degradation grade.

The invention provides a dynamic ad hoc network positioning method and system based on cooperation of multiple unmanned aerial vehicles, and belongs to the technical field of unmanned aerial vehicle communication and localization, and the method comprises the steps: firstly obtaining an arrival angle spectrum and bit error rate data of a communication link, and correcting a spatial covariance matrix through employing a weight factor mapped by the bit error rate; then, determining a relative azimuth vector through characteristic decomposition and orthogonal projection, and constructing an observation matrix; and then mapping the observation matrix to a geometric configuration constraint space to establish a topological cost function, and utilizing gradient iteration to update and solve a target configuration. And finally, completing coordinate conversion in combination with the reference position information to realize global positioning of the unmanned aerial vehicle cluster. The method can effectively improve the precision of unmanned aerial vehicle cluster cooperative positioning in a dynamic environment by introducing bit error rate weighted correction angle estimation and combining geometric manifold constraint optimization.

The present disclosure provides a method and system for determining a speech presence probability, a speech enhancement method and system, and a headphone. The speech presence probability and a speech absence probability in an iteration operation may be corrected by comparing an entropy of the speech presence probability and an entropy of a speech absence probability, such that a faster convergence speed and better convergence results may be obtained, thereby improving accuracy of an estimation of the speech presence probability and an accuracy of an estimation of a noise spatial covariance matrix, and then improving a speech enhancement effect of a minimum variance distortionless response (MVDR).

The invention relates to the technical field of seismic data interpretation, and particularly discloses a virtual well construction method and device based on seismic trace matching, and the method comprises the steps: carrying out the statistical analysis of logging curve data, and obtaining a spatial covariance matrix and a parameter covariance matrix; constructing a virtual well based on the spatial covariance matrix and the parameter covariance matrix; performing forward modeling on the virtual well and the known well to obtain a simulation record; and based on the well-side seismic trace and the simulation record, constructing an objective function, and obtaining a virtual well at a target position. The method comprises the following steps: calculating a space covariance matrix and a parameter covariance matrix by using data after logging statistical analysis, carrying out virtual well random simulation by solving a comprehensive covariance matrix of the first two items, carrying out seismic trace matching by using a deep network, obtaining a mapping relation between seismic data and logging parameters, and finally inputting the seismic trace into the deep network, so as to obtain the seismic data and the logging parameters. The virtual well at the target position is obtained, so that the well group formed by the real well and the virtual well is distributed more uniformly, and the modeling stability and the inversion precision are improved.

This invention discloses a Bayesian inversion extraction method for the equilibrium signal of wide-span altimeter data. The method involves acquiring and preprocessing sea surface height anomaly sequences from radar interferometers and nadir altimeters. A normalized sine square window function is applied for windowing, and multidimensional spatial averaging is used to estimate the one-dimensional wavenumber power spectrum. A piecewise power law-based equilibrium signal spectrum model and a noise spectrum model constrained by dynamic sea state are constructed, and the set of spectral parameters is extracted through logarithmic domain weighted least squares fitting. A set of spatial covariance matrices is constructed using cosine integral transform and Abelian forward and inverse transforms. A graphics processor is scheduled to perform batch matrix decomposition and singular fault-tolerant regularized inversion to solve for the posterior mean vector and posterior covariance matrix of the target equilibrium signal. Window fusion and index mapping are applied to fill the gaps in nadir observations. Geostrophic dynamics parameters are calculated, uncertainty quantification is performed based on the linear error propagation law, and the knowledge base is updated based on the exponential moving average algorithm. This invention achieves suppression of observation noise and physical filling of observation gaps, improving the adaptability of the inversion system to environmental changes while preserving non-Gaussian dynamic characteristics.

The application provides a positioning method and device, belongs to the technical field of sensing, and can be used for intelligent driving. The radar comprises a plurality of transmitting antennas and a plurality of receiving antennas for transmitting different carrier frequency signals. The method comprises the following steps: obtaining peak values of a plurality of two-dimensional transformation results, each two-dimensional transformation result being obtained by performing two-dimensional Fourier transformation on echo signals corresponding to a transmitting-receiving channel; performing eigenvalue decomposition on a covariance matrix to determine a noise subspace, each element in the covariance matrix being a peak value; constructing a spatial spectrum according to the noise subspace and a steering vector, the steering vector being determined according to the antenna configuration of the radar and the frequency of each transmitting signal, the steering vector being related to position information, and the position information comprising target distance and target direction of at least one target; and determining the position information, the position information making the spatial spectrum reach a maximum value. The position information is determined in a direction-distance joint estimation mode, and the positioning accuracy is improved.

The invention discloses a self-adaptive clustering method for unmanned aerial vehicle-mounted 4D radar point clouds, and belongs to the field of radar signal processing. The method comprises the steps of obtaining an original point cloud of an unmanned aerial vehicle 4D radar, and performing motion compensation and dynamic ROI filtering preprocessing; dBSCAN parameter self-adaption is driven through a radar physical model, the neighborhood radius and the minimum core point number are dynamically adjusted, and coarse clustering is carried out on point clouds; setting a clustering quality evaluation standard, calculating a spatial covariance matrix, a point cloud number and geometric distribution characteristics of a coarse clustering cluster, and screening out a to-be-refined cluster with an under-segmentation risk; after a to-be-refined cluster is screened, a heterogeneous graph model fusing space, speed and intensity features is constructed, and an RCS condition gating strategy is adopted; and carrying out recursive cutting by utilizing spectral clustering. According to the method, the problems of over-segmentation and under-segmentation caused by sparse point clouds along with distances and target RCS flicker are effectively solved, and the target sensing precision and robustness of the unmanned aerial vehicle in a complex scene are improved.

The invention relates to the technical field of attention mechanisms, and discloses an interrogation psychological sensitive point detection method based on a time sequence attention mechanism, which comprises the following steps of: processing microphone signals in a local time window through short-time Fourier transform to generate a space covariance matrix time sequence under a symmetrical positive definite manifold; extracting a relative displacement weighting kernel reflecting inter-frame time delay correlation from the output of the time sequence attention model; constructing a double-tangent vector based on affine invariant measurement, calculating a double-flow section curvature density index, and quantifying the geometric inconsistency of the double-flow section curvature density index and the double-flow section curvature density index; a weighted kernel is adjusted through index gating, a geometrically aligned prediction space covariance matrix is obtained through tangent space summation and exponential mapping, actual and prediction matrix residual errors are calculated through affine invariant measurement geodesic distance, and a single time sequence score is generated through weighting according to an energy proportion; and finally, a self-adaptive threshold value is constructed by using the absolute deviation of the graded median and a threshold value coefficient, and sensitive moments meeting conditions are screened, so that the detection accuracy and the scene adaptability are effectively improved.

The invention discloses a robust adaptive beam forming method based on distance unit sparse modeling, which comprises the following steps of: independently carrying out angle domain sparse modeling and processing on an array observation signal corresponding to each distance unit by taking a radar distance unit as a basic processing granularity; an angle domain sparse coefficient vector reflecting energy distribution in each angle direction is obtained through an angle domain over-complete guide dictionary, and an equivalent space covariance matrix is constructed by using the angle domain sparse coefficient vector and the angle domain over-complete guide dictionary instead of constructing the equivalent space covariance matrix based on a sample statistical meaning of an array receiving signal. An interference dominant subspace is extracted on the basis of an equivalent space covariance matrix, an optimal array beam forming weight vector is calculated in combination with target direction constraint, and weighted synthesis is carried out, so that under the conditions of strong interference, few snapshots and array mismatch, effective suppression of interference signals can still be realized, and accurate pointing of a beam main lobe in the target direction is kept.

The invention discloses a communication and sensing integrated multi-target direction of arrival estimation method, which comprises the following steps of: constructing a communication and sensing integrated system for describing a communication and sensing collaborative architecture between a first base station and a second base station, and constructing a reconfigurable intelligent reflecting surface signal receiving model and a second base station signal receiving model under the system; carrying out optimization processing on the second base station signal receiving model, constructing a positive semidefinite programming problem model based on atom norm minimization, and taking a minimization signal reconstruction error as an optimization target; carrying out decomposition solving on the matrix by adopting an alternating direction multiplier method of a dual-path self-adaptive penalty mechanism based on residual driving so as to output a Hermitian Toeplitz matrix; and constructing a reconstructed space covariance matrix based on the optimized Hermitian Toeplitz matrix, and performing spectrum peak search by adopting a multiple signal classification algorithm to finally obtain a multi-target direction of arrival estimation value. According to the method, a synergistic effect is formed in three aspects of interference suppression, robust estimation and low-complexity optimization, an efficient and feasible technical scheme is provided for high-precision and low-delay moving target positioning in a 6G sensing integrated system, and meanwhile, a lightweight sensing solution is provided for the Internet of Things with limited resources.

The invention discloses an underwater sound source localization model and method based on CNN-Transform and a double attention mechanism, and the model comprises a feature extraction module which is used for extracting the local features of the underwater sound source space covariance matrix tensor; the pixel-level channel attention module is used for re-calibrating the extracted local feature channel to obtain re-calibrated features; the Transform modeling module is used for introducing self-adaptive two-dimensional position coding into a Transform encoder based on the characteristics of the re-calibration, and capturing global context information by using the Transform encoder so as to obtain output characteristics; the dynamic large kernel space attention module captures an array space mode and fine features in positioning based on the output features to obtain a feature map; and the prediction module is used for estimating the sound source distance and depth based on the feature map. By adopting the model and the method, spectrum diffusion interference and feature degradation are cooperatively inhibited through double attention mechanisms, and a new scheme is provided for realizing high-precision sound source localization in a complex marine environment.

The invention discloses a multi-radiation source target direct tracking method, which is characterized in that a plurality of spatially distributed receiving arrays perform data acquisition on an unknown number of targets, each receiving array is composed of a plurality of array elements, and snapshot data are synchronously acquired and received, so that a receiving signal model under a multi-array collaborative observation condition is established. For the model, a plurality of spatial covariance matrixes are constructed, noise subspace information of the spatial covariance matrixes is extracted, and a MUSIC spatial spectrum is formed in combination with a subspace decomposition theory. According to the method, a spatial spectrum value is used as an observation likelihood function, and a generalized Labeled Multi-Bernoulli (GLMB) filtering framework is embedded, so that joint estimation of the number and the state of multiple targets is realized. And finally, performing propagation and weighted updating on the state particle set by combining a sequential Monte Carlo method to obtain the position of the target.

This application provides a dynamic ad hoc network positioning method and system based on multi-UAV cooperation, belonging to the field of UAV communication and positioning technology. First, this application acquires the arrival angle spectrum and bit error rate data of the communication link, and corrects the spatial covariance matrix using a weighting factor mapped by the bit error rate. Then, it determines the relative azimuth vector and constructs the observation matrix through eigenvalue decomposition and orthogonal projection. Next, it maps the observation matrix to a geometric configuration constraint space to establish a topological cost function, and uses gradient iteration to update and solve for the target configuration. Finally, it combines the reference position information to complete the coordinate transformation, achieving global positioning of the UAV swarm. This application can effectively improve the accuracy of UAV swarm cooperative positioning in dynamic environments by introducing bit error rate-weighted correction of angle estimation and combining it with geometric manifold constraint optimization.

The application provides a neural network-based marine wind field prediction method, and belongs to the technical field of marine wind field prediction.The application collects sparse marine observation data and establishes a spatial covariance matrix, converts the spatial covariance matrix into a graph structure, aggregates multi-hop neighborhood features by using a graph convolution network, models high-order feature interaction by combining a tensor decomposition algorithm to generate a gridded wind field, extracts spatiotemporal invariant features by using a bidirectional long short-term memory network encoder, directly maps future multi-step wind fields by using a multilayer perceptron predictor, and cooperates with a curriculum learning strategy and a neural ordinary differential equation boundary layer correction, so that the technical problem that sparse marine observation data is difficult to accurately reconstruct into a high-resolution gridded wind field is solved.

The invention discloses a homomorphic filtering-based distributed array de-reverberation sound source positioning method and system, and the method comprises the steps: synchronously collecting multi-channel reverberation acoustic signals based on a plurality of microphone arrays which are distributed; performing homomorphic filtering de-reverberation processing on each channel in the multi-channel reverberation acoustic signal so as to obtain a full-pass signal after de-reverberation; performing short-time Fourier transform on the full-pass signal after reverberation removal to obtain a time-frequency domain signal; calculating a spatial covariance matrix subjected to phase transformation weighting based on the time-frequency domain signal; constructing a steering vector based on a near-field propagation model; calculating steering response power by combining a spatial covariance matrix and the steering vector; and determining a sound source position based on the guiding response power. According to the method, homomorphic filtering and distributed array positioning are deeply fused, the positioning precision and robustness are remarkably improved, and an important technical support is provided for fault diagnosis and preventive maintenance of power equipment.

The invention relates to the technical field of new-generation mobile communication base station equipment and transmission, and discloses an RIS-MIMO signal transmission method based on adaptive beamforming, which comprises the following steps: a base station calculates a spatial covariance matrix according to uplink detection signals accumulated in a first preset time period, and executes subspace decomposition to extract slowly varying statistical characteristics; the base station retrieves a target code word based on a null space constraint criterion and drives a reconfigurable intelligent surface (RIS) to lock a phase so as to construct a static reflection channel; in the RIS locking period, the base station performs digital pre-coding on the data stream to be transmitted and transmits the data stream based on the equivalent channel state information containing the channel, and the method utilizes a dual-time-scale decoupling mechanism, configures the RIS through long-period statistical characteristics to construct a stable physical transmission boundary, and solves the problems of channel aging and pilot frequency overhead in a fast fading scene.

The invention relates to the technical field of intelligent man-machine interaction, in particular to an intelligent interaction method and system based on voiceprint recognition. The method comprises the following steps: acquiring a voice signal through a microphone array, framing the voice signal, performing real-time noise reduction by adopting an improved Wiener filtering algorithm, determining effective voice endpoints in combination with a double-threshold energy entropy detection method, and extracting voice segments containing voiceprint features; according to the method, the improved Wiener filtering algorithm based on the microphone array space covariance matrix and the dynamic threshold endpoint detection technology are introduced in the voiceprint collection preprocessing stage, the signal-to-noise ratio and the endpoint detection accuracy of the target voice signal in the complex environment are effectively improved, and high-reliability basic data are provided for subsequent feature modeling.

The invention relates to the technical field of industrial equipment nondestructive testing and fault diagnosis, and discloses an auxiliary machine stress wave state monitoring and diagnosis method and system based on multi-point acquisition, and the system uses a stress wave sensor array in non-collinear distribution to sense transient elastic wave signals generated by auxiliary machine equipment; the central processing unit constructs a parameter search vector containing a dynamic elastic modulus and an equivalent wall thickness, and performs frequency domain mapping on the signal; traversing the parameter search vector to generate a back propagation operator and reconstruct a wave field, and calculating a space covariance matrix and a space Shannon entropy value of the reconstructed wave field; and locking an optimal physical parameter group and a dominant propagation mode matched with the current working condition based on a minimum space entropy criterion. According to the method, the medium attribute change caused by temperature or corrosion can be adaptively corrected, the coordinate of the fault source and the wall thickness corrosion rate are accurately output, and the focusing precision and robustness of auxiliary machine fault diagnosis under the variable working condition are remarkably improved.

The invention relates to a two-dimensional DOA (Direction of Arrival) estimation method under an analog-digital hybrid array architecture, which comprises the following steps of: firstly, decoupling a two-dimensional angle estimation problem into two one-dimensional estimations by adopting an L-shaped antenna array and a partially connected hybrid analog-digital architecture; secondly, signal-to-noise ratio calculation is carried out through multi-sector scanning based on discrete Fourier transform, and rapid coarse estimation of the azimuth angle and the elevation angle is achieved; and then, reconstructing a spatial covariance matrix in a local angle interval corresponding to a coarse estimation result, and performing fine spectrum peak search by using a MUSIC algorithm to obtain a high-precision angle estimation value. And finally, by calculating the two-dimensional MUSIC frequency spectrum value of candidate angle pairing, the angle pairing problem between the subarrays is solved, and a final two-dimensional DOA estimation result is output. Through a multi-stage processing strategy of decoupling estimation, partition search and local refinement, high precision and high resolution under the condition of low signal-to-noise ratio are ensured while the calculation complexity is remarkably reduced.