28 results about "Residual matrix" patented technology

The invention discloses a training method of an image classification model for small sample incremental learning. The training method comprises the following steps: acquiring a pre-training visual encoder comprising a plurality of converter layers and a first classifier, and constructing a learnable prompt vector group to carry out reference stage iterative training; constructing a structure alignment branch parallel to the first classifier, and carrying out joint training by taking approaching the first classification output as a target to obtain a prompt vector group of structure alignment; extracting image features by using a prompt vector, constructing prototype matrixes of various categories, constructing a trainable low-rank residual matrix as residual compensation, forming a second classifier by the sum of the two matrixes, and training the second classifier; and finally outputting a visual encoder containing a prompt vector group and a second classifier for image classification in a reasoning stage. According to the method, the expressive power of the pre-trained visual language model can be fully utilized, and through low-rank structure alignment and continuous residual adaptation, a new category is efficiently learned while old-class knowledge is stably kept.

The invention relates to a direction-of-arrival estimation method based on a linear-nonlinear branch fusion neural network, and the method comprises the steps: obtaining an original signal of an antenna array, calculating a reference empirical covariance matrix based on the original signal, and obtaining an input tensor based on the reference empirical covariance matrix; based on the input tensor, processing by using a dual-path neural network to obtain a residual matrix; the dual-path neural network comprises a linear branch and a nonlinear branch; obtaining an enhanced covariance matrix based on the residual matrix and the reference empirical covariance matrix; and based on the enhanced covariance matrix, obtaining a DOA estimation value by using a Root-MUSIC algorithm. Compared with the prior art, the direction-of-arrival estimation method provided by the invention has the advantages of a traditional subspace algorithm and a neural network participation type method.

This invention relates to the field of intelligent rolling technology, and discloses an intelligent rolling method and system for aluminum-based bimetallic bearing materials based on three-roll differential rolling. The method includes: acquiring real-time multidimensional variable data of three-roll differential rolling; obtaining principal component eigenvectors by dimensionality reduction after normalization and principal component analysis; constructing a reconstruction matrix and calculating the squared Euclidean norm of the residual matrix as the squared prediction error value; analyzing the contribution if the error exceeds a threshold, locating the precise coordinates of the profile deviation; maintaining the original parameters if the error does not exceed a threshold; generating adjustment parameters and compensation commands through a roll elastic deformation model based on the mill geometric data, and outputting them to the drive system; calculating the dynamic following error, performing secondary compensation for exceeding the threshold, and finally obtaining optimized rolling parameters. This method can achieve precise positioning and targeted compensation of rolling profile deviation, ensuring stable product quality.

The application discloses a non-line-of-sight identification method and device based on subset band signed residual inspection and a medium. The application comprises grouping positioning measurement data, intermediate estimation of the grouped data, calculation of a complete initial subset band signed residual matrix by using the intermediate estimation value, detection of a non-line-of-sight candidate hypothesis based on the initial subset band signed residual matrix, and establishment of a final non-line-of-sight base station combination from the candidate hypothesis. The application uses a subset band signed residual matrix, can effectively reduce the influence of front-end estimation on residual values, and improves the identification accuracy of an existing algorithm. The ML inspection proposed by the application unifies the selection process and criteria in multiple candidate hypothesis cases, can make a choice according to a maximum likelihood measurement obtained by strict calculation, and improves the selection accuracy of the candidate hypothesis.

The invention belongs to the technical field of oil and gas field development engineering, and particularly relates to a conglomerate reservoir horizontal well perforation completion discrete element modeling method. The method comprises the following steps: extracting a conglomerate contour of a target area; taking the conglomerate contour of the extracted target area as a constraint boundary, generating a plurality of discrete small ball particles to fill the interior of the conglomerate contour and the residual matrix area, and further generating a conglomerate discrete element model; the method comprises the following steps: acquiring field perforation well completion data of a horizontal well in a target area, and converting the field perforation data to generate a horizontal well shaft wall unit containing perforations; the field perforation completion data at least comprise perforation depth, perforation aperture, perforation density, wellbore diameter and wellbore length parameters. Through the conglomerate reservoir perforation horizontal well fracturing simulation technology, the limitation of a traditional homogeneous model is broken through, and accurate reduction of the heterogeneity of a real reservoir is achieved.

The invention relates to the technical field of data processing, in particular to a 3D printing material defect detection method and system based on multi-sensor fusion. The method comprises the steps of capturing thermal radiation flux and a pose change track, extracting thermodynamic divergence and a transient motion vector, constructing a spatial delay compensation coefficient by combining a heat exchange dissipation coefficient, generating a priori estimation set by a feedforward correction state prediction matrix, constructing a divergence suppression factor according to a transient temperature change rate, and multiplying the divergence suppression factor into a covariance prediction link. Filtering to generate a posterior estimation set, extracting a reflectivity optical gradient for cross-domain verification, mapping an optical reference and calculating a residual matrix, superposing a high-frequency residual to the posterior estimation set, generating a defect reconstruction set, and analyzing the set to output defect space coordinates and morphology distortion information. According to the invention, the printing material defect detection precision is obviously improved.

The present disclosure relates to a method for predicting the enthalpy of a phase change material, a near-infrared spectrum of a phase change material sample is subjected to a partial least squares regression analysis with a standard enthalpy to establish a first calibration model; then a spectral fitting residual matrix is selected according to the construction process of the first calibration model, the spectral fitting residual matrix and the enthalpy matrix are combined, and are used as input signals and teacher signals of an extreme learning machine respectively to obtain a second calibration model; through the combined application of the first calibration model and the second calibration model, the linear relationship and the nonlinear relationship between the near-infrared spectrum of the phase change material sample and the enthalpy can be comprehensively considered, the enthalpy prediction capability in a wide range is achieved, and the model prediction accuracy is improved.

The invention provides a structured sparse compensation method and system based on residual energy perception. The method comprises the following steps: calculating a quantized residual matrix of a weight in a quantized backbone network; constructing a compensation branch with a regular sparse structure based on the energy distribution characteristics of the quantized residual matrix; and integrating the compensation branch and the quantization backbone network in a parallel connection mode, and realizing compensation of an output error introduced by weight quantization by superposing output of the compensation branch and the quantization backbone network in a reasoning process. The structured sparse compensation branch has a high-rank expression capability, does not depend on global low-rank hypothesis, constructs a regularly arranged sparse substructure based on residual energy distribution, effectively approaches a high-frequency error, remarkably improves the detail recovery capability, and solves the problems of blurring and distortion of a generated image.

The application provides a kind of tank sealing quick detection method, system and storage medium, including obtaining and preprocessing the original signal data of tank detection, obtain input sample data;Initialize width learning model, build initial feature node set and calculate initial output weight matrix;Node increment iteration is carried out until model AIC value does not decline for two times in succession or increment node number reaches upper limit;Each iteration includes: calculate the residual matrix between prediction and label, determine the number of candidate feature nodes to be generated according to F norm, and generate weight to build candidate pool using Sobol sequence;Calculate the Pearson correlation coefficient of each candidate node and residual, select the node with the largest absolute value as the incremental node and incorporate into the model, and update the output weight matrix using block matrix inversion method;When the number of new nodes reaches the pruning period, set the pruning threshold, and remove the feature nodes with weight norm less than the threshold;The model obtained after iteration termination is used to discriminate the tank to be tested.

The invention relates to a rainfall data analysis method based on multi-modal data fusion and space-time correlation perception, and the method comprises the steps: collecting historical non-blocking operation and maintenance data of a rain gauge in a target region in a set time period through a system, and storing the historical non-blocking operation and maintenance data in a computer system in a partitioned manner to construct an equipment historical baseline library; the system receives real-time multi-mode monitoring data of a preset number of rain gauges in a target area to construct a real-time monitoring data set, converts historical non-blocking operation and maintenance data and time sequence data of the real-time monitoring data set into scores capable of being quantized in a unified mode according to time step column values, and determines analysis points and peripheral points in the target area. And calculating a score residual error between each time step column value and a baseline midpoint value in a historical baseline library, generating a two-dimensional baseline residual error matrix, calculating a dynamic deviation index and a state continuity parameter, constructing a blockage judgment function, and combining an operation result with a preset threshold value to complete blockage judgment. According to the method, full-process automation can be realized, and efficient analysis is realized.

Provided is a method of predicting a condition of a patient, including decomposing an input matrix to produce a residual matrix, the input matrix representing patients and expression levels of genes of the patients, training a classifier by using health condition values the patients as learning criteria and inputting the residual matrix into the classifier, and obtaining a value for predicting a health condition of a first patient, by inputting a first input matrix including the expression levels of genes of the first patient into the classifier, the first residual matrix being a residual matrix obtained from the first input matrix by using the predetermined algorithm.

The invention relates to the technical field of satellite data processing, and provides a gravity satellite vacancy data interpolation method and device, and the method comprises the steps: obtaining a plurality of pieces of historical satellite gravity data and corresponding multi-source hydrometeorological auxiliary data in a preset time period; performing empirical orthogonal function decomposition on the plurality of historical satellite gravity data to obtain a plurality of spatial modes; acquiring a plurality of continuous time coefficient sequences according to the plurality of historical satellite gravity data and the plurality of spatial modes; constructing a preliminary reconstruction field according to the plurality of spatial modals and the plurality of continuous time coefficient sequences; and determining a to-be-predicted residual error matrix according to the plurality of historical satellite gravity data and the preliminary reconstruction field, and obtaining a corresponding to-be-predicted residual error coding matrix. And according to the preliminary reconstruction field, the multi-source hydro meteorological auxiliary data, the to-be-predicted residual matrix and the to-be-predicted residual coding matrix, predicting a prediction result value of the to-be-interpolated satellite gravity data through a prediction model. And according to the prediction result value and the preliminary reconstruction field, performing interpolation processing on the to-be-interpolated satellite gravity data.

The application discloses a kind of earth-rock dam stress deformation field time series prediction method and system based on finite element prior and monitoring data assimilation, belong to the technical field of water conservancy and hydropower engineering, method includes: multiple types of monitoring data and complete finite element field establish unified physical space registration relationship, and the prior stress deformation field of earth-rock dam is encoded into prior latent variable;Prior stress deformation field is mapped to sparse field monitoring space, and compared with field monitoring data matrix to construct monitoring residual matrix and feature extraction, correct prior latent variable in low-dimensional latent space based on sparse monitoring correction feature, and reconstruct as finite element complete stress deformation field after monitoring assimilation;Through autoregressive propagation model, online assimilation and rolling prediction are carried out.The application realizes dynamic correction and future state prediction of finite element full field prior under sparse monitoring condition, improves the spatial integrity, real-time performance and accuracy of safety state evaluation of earth-rock dam construction period and initial operation period, and has significant practicability and wide application prospect.

The application relates to a linear-nonlinear branch fusion neural network-based direction of arrival estimation method, which comprises the following steps: obtaining original signals of an antenna array, calculating a reference empirical covariance matrix based on the original signals, and obtaining an input tensor based on the reference empirical covariance matrix; based on the input tensor, a residual matrix is obtained by using a double-path neural network; the double-path neural network comprises a linear branch and a nonlinear branch; an enhanced covariance matrix is obtained based on the residual matrix and the reference empirical covariance matrix; and a DOA estimation value is obtained by using a Root-MUSIC algorithm based on the enhanced covariance matrix. Compared with the prior art, the application provides a direction of arrival estimation method which can have the advantages of both a traditional subspace algorithm and a neural network participation type method.

The invention provides a switch cabinet fault diagnosis method and device, electronic equipment and a storage medium, and belongs to the technical field of fault diagnosis, and the method comprises the steps: collecting a current time sequence signal of a switch cabinet, and building an observation matrix of the current time sequence signal; constructing a null space projection operator orthogonal to the healthy current time sequence signal; projecting the observation matrix to a null space by using a null space projection operator, and separating a residual matrix representing an abnormal current time sequence signal; according to the residual matrix, whether the switch cabinet has a fault is judged, and if yes, a fault dictionary matrix describing the fault category of the switch cabinet is constructed; otherwise, returning the current time sequence signal, and establishing an observation matrix of the current time sequence signal; the residual matrix and the fault dictionary matrix are combined, and the fault similarity of the current time sequence signals and different switch cabinet fault categories is calculated; and outputting the switch cabinet fault category corresponding to the maximum fault similarity as a switch cabinet fault diagnosis result. The accuracy and efficiency of fault diagnosis of the switch cabinet can be improved.

The invention provides a data processing method and device. The data processing method comprises the steps of obtaining a residual matrix of a target layer based on an original weight matrix for the target layer in a neural network and a pseudo-quantization weight matrix corresponding to the original weight matrix; obtaining an activation vector of the target layer based on the activation data for the target layer; obtaining a target channel index based on the residual matrix and the activation vector; and determining a target weight channel in the original weight matrix based on the target channel index.

The invention relates to the field of drug packaging material barrier property simulation and data processing, in particular to a method for calculating and processing drug packaging bag barrier property twinborn simulation data, which comprises the following steps: acquiring original grid topological data, time sequence permeation flux data, a defect-free diffusion model and a defect feature knowledge base of a multilayer medium structure; generating an ideal blocking reference field, and mapping the ideal blocking reference field into a reference flux matrix; calling a defect parameterization operator to generate a theoretical defect state and calculating a theoretical residual matrix; calculating a real residual matrix; and performing space-time similarity calculation on the real residual matrix and the theoretical residual matrix to generate a defect confidence result, retaining real defect region data, performing elimination or replacement processing on numerical noise region data, and outputting corrected simulation data. Updating defect parameterization operator parameters or similarity judgment threshold value groups on the basis of a plurality of continuous processing cycles; according to the method, numerical noise can be eliminated while real tiny defects are reserved.

This invention provides a real-time monitoring and early warning method and device for crane hooks based on multi-source data fusion, relating to the field of artificial intelligence technology. It adaptively segments the real-time monitoring sequence based on the velocity envelope fluctuation intensity of the lifting height sequence, avoiding feature aliasing caused by start-stop, speed change, and other operations. Based on the sampling point differential sequence, data from each channel is coupled, quantifying the intensity of synchronous changes between load, lifting, and oscillation into comparable anomaly sensitivity indicators. Based on the difference in motion trends between the coupling characteristics and the aforementioned segmented characteristics, a trend matrix and a residual matrix are decomposed. This allows the separation of the operator's active operational intent from the normal inertial motion of the load, enabling the residual matrix to centrally reflect abnormal fluctuations such as mechanical clearance and elastic oscillations. The model can effectively distinguish between normal fluctuations and early hidden dangers under strong dynamic and interference environments, achieving adaptive, high-sensitivity, and low-false-alarm-rate real-time anomaly identification, ensuring the inherent safety of lifting operations.

The application discloses an asphalt mixture-moisture characteristic curve model and a parameter calculation method, and the model and the parameter calculation method comprise the following steps: 1, preparing an asphalt mixture core sample; 2, determining the initial dry mass and porosity of the asphalt mixture core sample by using a vacuum saturation method, and checking the parallelism of parallel core samples; 3, measuring the matrix suction ψ of the parallel core samples of the asphalt mixture m and a data pair corresponding to the water saturation S; 4, substituting a series of experimental data of the matrix suction ψ m and the water saturation S into an initial model of the asphalt mixture-moisture characteristic curve, and fitting to obtain the asphalt mixture-moisture characteristic curve; 5-7, bringing the determined effective parameters into the initial model; 8, deriving the effective model; 9, calculating the slope of the zero matrix suction point; and 10, calculating a residual matrix suction parameter. The asphalt mixture-moisture characteristic curve mathematical model provided by the application is continuous and monotonically decreasing in the range from zero to infinity of the matrix suction, and is convenient for calculating the unsaturated parameters.

The application provides a chemical equipment variable condition implicit fault diagnosis method, device and equipment, comprising: obtaining a multivariate time series matrix of multiple variables, separating the trend value in the multivariate time series matrix to obtain a trend matrix; obtaining a residual matrix according to the multivariate time series matrix and the trend matrix; obtaining a slow feature projection matrix according to the trend matrix; obtaining a slow feature matrix according to the trend matrix and the slow feature projection matrix; iteratively performing: optimizing the trend matrix while keeping the slow feature projection matrix unchanged to obtain a new trend matrix, and then obtaining a new residual matrix, a new slow feature projection matrix and a new slow feature matrix, obtaining a function value of a preset target function, until the function change value of the preset target function is less than a preset function change value. When an abnormal alarm event is triggered, the contribution degree value of each variable to the abnormal alarm event is obtained; wherein the contribution degree value is used for fault diagnosis. The accuracy of the chemical equipment variable condition implicit fault diagnosis is improved.

The application discloses a dynamic expansion fault detection method based on a regression relation and relates to the technical field of fault detection engineering. The method comprises the following steps: obtaining a predictable dynamic information matrix according to an input data matrix and an output data matrix of a standard missile; obtaining a main score matrix and a number of principal components related to output according to the predictable dynamic information matrix; obtaining an input load matrix according to the main score matrix and a dynamic input information matrix; obtaining an input residual matrix according to the main score matrix and the input load matrix; obtaining a number of principal components irrelevant to output according to the input residual matrix; obtaining an input control limit and an output control limit according to the number of principal components; obtaining a statistic quantity related to output and a statistic quantity irrelevant to output according to an input data matrix and the input load matrix of a to-be-detected missile; and determining a fault detection result according to the statistic quantity related to output, the input control limit, the statistic quantity irrelevant to output and the output control limit. The application can reduce a false alarm rate and improve detection precision.

This invention relates to the field of ultrasonic testing technology, specifically providing a method and system for detecting internal defects in the layered structure of an insulating tie rod. The method includes: acquiring the echo amplitude of the insulating tie rod under different incident angles and constructing an actual detection matrix based on the corresponding sampling depth; performing dynamic structural noise cancellation based on the actual detection matrix and a reference template matrix to obtain a residual matrix, wherein the reference template matrix is ​​constructed based on the echo amplitude of an insulating tie rod sample without internal defects; comparing the residual amplitude of each sampling point in the residual matrix with a preset threshold to obtain candidate defect points; and obtaining a defect determination result based on the residual amplitude change and angle width of the candidate defect points through corresponding threshold comparisons. This addresses the problem in related technologies where ultrasonic testing of insulating tie rods cannot effectively distinguish between defect echoes and interlayer echoes, leading to poor defect detection results.

This invention belongs to the field of intelligent early warning technology, specifically relating to an intelligent early warning method for a DMF waste liquid purification and recovery control platform. The method includes: performing principal component analysis on long-term historical data to construct the principal component transformation matrix of a static benchmark model; for the time to be diagnosed: calculating the reconstructed value using the static benchmark model to obtain the residual vector; performing eigenvalue decomposition on the covariance matrix of the residual matrix of the sliding time window; calculating the drift coherence index based on the distribution of drift eigenvalues; modulating the drift principal component vector of the sliding time window using the residual vector to obtain a drift compensation vector; superimposing this vector with the reconstructed value of the real-time data vector at the time to be diagnosed to obtain an adaptive reconstructed value for the time to be diagnosed, used to calculate the reconstructed error; comparing this value with a fault alarm threshold to determine whether a fault exists at the time to be diagnosed and issuing an early warning. This invention improves the accuracy and robustness of early warning.

The invention discloses a multi-source data fusion network security knowledge model construction method, and relates to the technical field of network security data modeling and analysis, and the method comprises the steps: collecting multi-source data, carrying out the preprocessing, and generating a stable matrix; performing sequence prediction based on a stationary matrix, constructing a residual matrix and a completely undirected graph, calculating a Pearson's correlation coefficient between nodes to perform significance judgment to obtain a candidate set, and calculating an average correlation score of the nodes based on the candidate set to perform node interception to form an interception set; through a dynamic threshold adjustment mechanism, an exponential decay weighting method, fusion residual relevancy, a comprehensive similarity matrix, hierarchical clustering and a direction consistency index, a security knowledge cluster which is clear in structure and reasonable in hierarchy can be formed, so that the interpretability and the expandability of a network security knowledge model are remarkably improved.

The invention discloses a time sequence image-based power distribution network anomaly positioning method, device and equipment and a medium, and the method comprises the steps: obtaining measurement data collected from a power distribution network by a plurality of different sensors at a plurality of time points, and carrying out the preprocessing of the measurement data, and generating multivariable time sequence data; respectively converting the time sequence data corresponding to each time point in the multivariable time sequence data into a two-dimensional time sequence image; calling the target generative adversarial network, generating a prediction image corresponding to each two-dimensional time sequence image, and calculating reconstruction loss and feature loss corresponding to each two-dimensional time sequence image; weighting the feature loss according to a preset weighting coefficient, and superposing the reconstruction loss to obtain an abnormal score corresponding to each two-dimensional time sequence image; and if the abnormal score exceeds a preset abnormal threshold value, calculating a residual matrix between the two-dimensional time sequence image and the predicted image, and positioning an abnormal position according to the residual matrix. Therefore, high-precision detection of abnormal data is provided under the condition of fusing multiple data modalities.

This invention discloses an artificial intelligence-based method, equipment, and medium for monitoring energy consumption in tag production, relating to the field of production monitoring technology. The method includes: collecting tag production data, calibrating it, and defining a set of energy consumption monitoring nodes; defining anomaly factors and screening a set of steady-state second points; selecting energy consumption monitoring variables based on the anomaly factors, constructing anomaly scores for each anomaly factor, and converting them into activation vectors to generate state indices; dividing the data into windows and obtaining baseline sampling sets for each node to establish a window residual matrix; calculating the anomaly source intensity to establish a hypothetical observation model; processing the node residuals within the window and outputting multi-dimensional matrix features; constructing a parallel three-branch network to analyze the multi-dimensional matrix features, obtaining fused features, and mapping them to state probabilities and power predictions; and triggering early warnings based on the state probabilities output by the parallel three-branch network. This invention effectively improves the accuracy and timeliness of anomaly analysis in the tag production process.

The invention relates to the technical field of detection, in particular to a wave velocity field tomography method based on relative travel time data, and the method comprises the steps: S1, obtaining the original arrival time data of a vibration signal; s2, processing the original arrival time data of the vibration signal to obtain an actual measurement conversion arrival time vector; s3, an objective function is constructed based on the difference between the minimized observation data and the forward model prediction data, the objective function comprises a travel time residual item and a Tikhonov regularization item, and the weight of travel time residual convergence and model data smoothing is adjusted through a weighting factor; a Levenberg-Marquardt strategy is introduced to ensure the convergence of the inversion calculation; s2, solving partial derivatives of slowness adjustment vectors for the target function to obtain a regular equation, replacing a travel time residual matrix in the regular equation with the actually-measured conversion-to-time vectors in the S2, solving the replaced regular equation to obtain slowness adjustment values, and performing inversion to obtain the internal wave velocity field of the rock mass. And wave velocity field inversion can be completed only by using the relative arrival time of the signals received by the receiving points.

The application discloses a new energy power facility detection data management and analysis platform, comprising: a multi-source heterogeneous high-frequency synchronous induction system, each sensor node of which is equipped with a Beidou timing module, so that sampling error is controlled within 20ns by taking a second pulse as a sampling trigger source and adding a time stamp; an environment-driven dynamic reference flow form construction system for extracting low-dimensional embedding coordinates of environment characteristics through nonlinear flow form learning and constructing an ideal output hyper surface changing with the environment; and an endogenous performance deviation decoupling extraction system for calculating a residual matrix of real-time operation characteristics and ideal operation states and adopting an independent component analysis algorithm to decouple and extract an endogenous attenuation characteristic vector representing equipment intrinsic performance attenuation from the residual matrix. The application changes the monitoring reference from a fixed threshold to a dynamic baseline fluctuating with the environment, effectively separates environmental interference and equipment real performance attenuation, and improves diagnostic accuracy and preventive maintenance capability.