55 results about "Residual matrix" patented technology

The invention relates to the technical field of optical fiber sensing and fault diagnosis, in particular to an optical fiber gyroscope fault monitoring method and system. The method comprises the following steps: extracting four-dimensional characteristics of phase difference, light intensity fluctuation, polarization state drift and temperature drift from an output signal of an optical fiber gyroscope, and constructing a high-dimensional characteristic matrix after unifying a time reference; establishing a dynamic prediction model based on historical data to generate a residual matrix, and orthogonally separating the residual into an internal structure abnormal signal and an environment disturbance signal through covariance characteristic decomposition; mapping the structure residual error sequence into a topological graph, calculating an evolution index in real time, and capturing a fault gradient trend in combination with sliding window gradient analysis; a multi-dimensional vector is constructed by fusing topological features, an adaptive classifier of an online Gaussian mixture model is adopted to identify a fault mode and quantify a health index, and meanwhile, a health measurement result is fed back to a prediction model and topological analysis parameters to realize closed-loop optimization. According to the invention, full-process adaptive monitoring from anomaly detection to health measurement is realized.

The invention relates to a direct current charging pile fault diagnosis and repair method and device and a storage medium. The method comprises the following steps: carrying out Hilbert transformation on voltage, current and temperature parameters of a charging pile to obtain envelope data of a multi-channel analysis signal; performing frequency domain conversion and maximum posterior probability matrix decomposition on the envelope data to obtain a frequency spectrum feature vector; inputting the spectrum feature vector into a dual-control law observer with a continuous item and a discontinuous item for state tracking to obtain a system residual matrix; performing threshold detection and accumulative error integral operation on the system residual matrix to obtain fault feature data; and performing lambda consensus iterative operation on the fault feature data to obtain a control parameter compensation scheme. According to the method, the recognition accuracy of fault features is improved, and the optimality and reliability of a repair strategy are ensured.

The invention belongs to the technical field of intelligent early warning, and particularly relates to an intelligent early warning method for a DMF waste liquid purification and recovery control platform, and the method comprises the steps: carrying out the principal component analysis of long-period historical data, and constructing a principal component transformation matrix of a static reference model; for a moment to be diagnosed, calculating a reconstruction value by using the static reference model to obtain a residual vector, carrying out eigenvalue decomposition on a covariance matrix of a residual matrix of a sliding time window, calculating a drift coherence index according to the distribution of drift eigenvalues, modulating a drift principal component vector of the sliding time window by combining the residual vector, and carrying out diagnosis on the moment to be diagnosed; and obtaining a drift compensation vector, superposing the drift compensation vector with a reconstruction value of a real-time data vector at a to-be-diagnosed moment to obtain an adaptive reconstruction value at the to-be-diagnosed moment, calculating a reconstruction error, comparing the reconstruction error with a fault alarm threshold, judging whether a fault exists at the to-be-diagnosed moment, and performing early warning. According to the invention, the early warning accuracy and robustness are improved.

An online inference platform may obtain k past feature matrices, each feature matrix being associated with a set of nodes and, for each node, a feature vector (set of node measurements) representing operation of a cyber-physical asset. The k past feature matrices and an adjacency matrix are provided to a trained, structurally-aware normality model which creates a predicted feature matrix. A residual matrix Rt is obtained by calculating a difference between the predicted feature matrix for time t and a measured feature matrix for time t, and an asset detection decision (e.g., "normal" or "abnormal") is obtained by thresholding the residual matrix Rt. The online inference platform may also perform a localization process when the asset detection decision is "abnormal." An offline model training platform may create the normality model based on historical feature matrices of the asset and the adjacency matrix.

The invention discloses a mixed DNA sample donor proportion analysis method based on STR typing and residual matrix analysis, which comprises the following steps: firstly, calculating the minimum donor number in a sample through a maximum allele counting method, further fitting the genotype distribution of the sample through an optimization algorithm, comparing the difference between the predicted peak height and the actually measured peak height, and determining the donor proportion of the sample according to the difference between the predicted peak height and the actually measured peak height. Therefore, the proportion of each donor in the mixed DNA sample is accurately estimated. Experimental results show that when the method is used for processing a multi-allele mixed DNA sample, estimation errors existing in a traditional method can be effectively reduced, and particularly, high accuracy is shown in complex sample analysis. By establishing a linear model and combining an optimization algorithm, the method not only provides high-precision technical support for DNA analysis in forensic science, but also provides a feasible solution for quantitative analysis of mixed DNA samples in other fields.

The invention provides a method for calibrating hyperspectral reflectivity data of field tea ridges, which comprises the following steps of: acquiring outdoor hyperspectral data and indoor hyperspectral data of the field tea ridges, and performing cubic B spline fitting on the indoor hyperspectral data; according to the wave band position and number of the outdoor hyperspectral data, the reflectivity of the corresponding wave band position is obtained from the indoor spectral reflectivity curve, and an indoor standard reflectivity curve is obtained. Searching the modulation coefficient of the k-th iteration under the condition that the constraint function obtains the minimum value; calculating a correction factor function according to the modulation coefficient of the kth iteration; substituting the training set and the correction factor function into a calibration formula, and substituting the verification set, the kth environmental interference error and the correction factor function into the calibration formula; and if the kth residual matrix meets the preset condition, iteration is stopped. According to the method, the optimal modulation coefficient can be searched, so that the outdoor spectral reflectivity can be calibrated adaptively, and the outdoor hyperspectral data of the field tea ridge can be measured more accurately.

The invention relates to the technical field of safety reinforcement learning, and discloses a deep Koopman-based robot reinforcement learning safety controller online updating method, which comprises the steps of collecting a trajectory state of random input control in simulation, training a deep Koopman neural network, and obtaining a corresponding lifting function and an evolution matrix; carrying out dimension reduction processing on the model by adopting an intrinsic orthogonal decomposition method to obtain a projection matrix and a new nominal model; carrying out reinforcement learning strategy migration in the real machine, obtaining an observation error according to the nominal model and the current observation state in interaction, and training an online update network to obtain a residual matrix; the nominal model and the residual model are combined to serve as model constraints of model prediction control, and safety control input is obtained. According to the method, the safety controller can be updated online, the reinforcement learning safety guarantee performance is improved, and the method is suitable for complex situations that model differences exist in the process of migrating a reinforcement learning strategy from simulation to a real machine, and disturbed dynamic environments exist in the physical world and the like.

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 embodiment of the invention provides a power supply state fault positioning method and system, electronic equipment and a storage medium, and belongs to the technical field of fault positioning. The method comprises the following steps: carrying out residual calculation according to an original time domain electromagnetic signal and a reference frequency spectrum template to obtain a frequency spectrum residual matrix; and performing cross-frame disturbance integral scoring on the spectrum residual vector according to the integral residual scoring function to obtain disturbance scoring data. And obtaining a disturbance event set according to the disturbance score data and the disturbance threshold. And fusing the disturbance event set and the operation log to obtain a state event sequence. And constructing a state semantic tree according to the state event sequence and calculating an edge weight. And scoring each path according to the path scoring function and the edge weight to obtain path scoring data. And selecting a path corresponding to the maximum path score data as a main cause path, wherein a head node of the path is a fault root cause node. And entering a preset execution output process according to the main cause path and the fault root cause node. And the accuracy of power supply fault positioning is improved.

The invention relates to a method for predicting an enthalpy value of a phase change material, and the method comprises the steps: carrying out the partial least square regression analysis of a near infrared spectrum of a phase change material sample and a standard enthalpy value, and building a first correction model; then selecting a spectrum fitting residual matrix according to the construction process of the first correction model, and taking the spectrum fitting residual matrix and the enthalpy value matrix as an input signal and a teacher signal of an extreme learning machine respectively to obtain a second correction model; through the combined application of the first correction model and the second correction model, the linear relation and the nonlinear relation between the near infrared spectrum and the enthalpy value of the phase change material sample can be comprehensively considered, the enthalpy value prediction capacity in a wide range is achieved, and the model prediction precision is improved.

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.

The application provides a channel estimation method of a visible light communication system, and relates to the technical field of channel estimation. The method comprises the following steps: obtaining a pilot observation matrix of a receiving end; calculating an initial channel estimation by using an LMMSE estimator; splicing the initial channel estimation and a noisy state of a diffusion time step to obtain a joint input tensor; inputting the joint input tensor into a pre-trained LMRDNet, recovering a predicted residual matrix from the noisy state through a reverse diffusion process; and calculating refined channel state information according to the initial channel estimation and the predicted residual matrix. Through the fusion of an LMMSE statistical priori and a customized lightweight diffusion model, the application realizes high-precision channel estimation under the condition of limited pilots and low signal-to-noise ratio.

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 invention discloses a sprouting vegetable yield prediction method and system based on artificial intelligence. The method comprises the steps of data collection, data analysis, yield dynamic prediction model construction, prediction model optimization and yield improvement optimization. The invention belongs to the technical field of intelligent agriculture, and particularly relates to a sprouting vegetable yield prediction method and system based on artificial intelligence, and the method employs a differential equation architecture and an attenuation kernel function to quantify a dynamic coupling relation between an environment factor and a growth process, employs a recursive least square method with a forgetting factor to achieve the online updating of parameters, and achieves the prediction of the yield of sprouting vegetables. An innovative dynamic adaptation index quantifies the response capability of the system to emergencies; sHAP value quantification is adopted to realize dynamic sorting of contribution degrees of multiple environmental factors, and a real-time anomaly detection system is constructed through a standardized residual matrix.

The application relates to the technical field of safety reinforcement learning, and discloses a robot reinforcement learning safety controller online updating method based on deep Koopman, which comprises the following steps: collecting trajectory states of random input control in simulation, training a deep Koopman neural network, obtaining corresponding promotion functions and evolution matrices, adopting an intrinsic orthogonal decomposition method to perform dimension reduction processing on the model to obtain a projection matrix and a new nominal model, performing reinforcement learning strategy migration in a real machine, obtaining an observation error according to the nominal model and a current observation state in interaction, training an online updating network to obtain a residual matrix, and combining the nominal model and the residual model as model constraints of model predictive control to obtain a safety control input. The application can update the safety controller online, improve the safety guarantee performance of reinforcement learning, and is suitable for complex scenes such as model differences in the process of migrating the reinforcement learning strategy from simulation to a real machine and dynamic environments with disturbances in the physical world.

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 invention provides an observation station system error correction method and system, computer equipment and a storage medium, and belongs to the field of spaceflight measurement and control, and the method comprises the steps: respectively obtaining observation matrixes of a laser ranging station SLR and a deep space network observation station, building observation equations of laser ranging and deep space network ranging and speed measurement, building an orbital motion equation for a satellite, and obtaining an orbital motion equation; superposing the SLR observation equation, the UXB observation equation and the orbital motion equation according to weights to construct a normal equation set; calculating theoretical prediction matrixes of the SLR and a deep space network observation station by combining an orbital motion equation, the SLR and an observation equation of the deep space network, calculating respective residual matrixes according to the difference between the prediction matrixes and the observation matrixes, and solving to-be-estimated parameter correction of an equation set by using a least square method; and comparing the orbit correction with a threshold value, obtaining a final to-be-estimated parameter correction through multiple iterations, and calibrating and compensating the measurement process of the deep space network tracking station by using the to-be-estimated parameter correction, thereby improving the satellite orbit determination precision.

The present invention discloses a method for analyzing the proportion of donors in mixed DNA samples based on STR typing and residual matrix analysis. First, the minimum number of donors in the sample is calculated by the maximum allele counting method, and then the genotype distribution of the sample is fitted by an optimization algorithm. The difference between the predicted peak height and the actual measured peak height is compared to accurately estimate the proportion of each donor in the mixed DNA sample. Experimental results show that the present invention can effectively reduce the estimation error existing in traditional methods when processing multi-allele mixed DNA samples, especially showing high accuracy in complex sample analysis. By establishing a linear model and combining it with an optimization algorithm, the present invention not only provides high-precision technical support for DNA analysis in forensic science, but also provides a practical solution for the quantitative analysis of mixed DNA samples in other fields.

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.

A spacecraft fault diagnosis method based on a probability invariant space comprises the following steps: firstly, establishing a discrete state space model of a spacecraft system; then, injecting a fault mode to obtain estimated observation data; eliminating the state quantity of unknown probability distribution according to the equivalent space model, obtaining a residual matrix of the estimated observed quantity, and calculating an estimated mean value and a covariance matrix in each fault mode; then, when the spacecraft operates in orbit, obtaining a mean value and a covariance matrix of an in-orbit observed quantity residual matrix according to the equivalent space model; then, according to the estimated mean value, the covariance matrix, the on-orbit mean value and the covariance matrix, the K-L divergence of the probability distribution of the on-orbit operation data and the estimated data of each fault mode is obtained; and finally, according to the K-L divergence, identifying the fault mode with the closest probability density during in-orbit operation of the spacecraft to complete fault diagnosis. When the method is applied to on-orbit fault diagnosis of the spacecraft, the complexity of a fault diagnosis algorithm is reduced only by comparing probability distribution differences of observation data.

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 medical data analysis, in particular to an immune marker analysis system based on machine learning, which comprises an entropy density rheological module, a boundary dynamic labeling module, a collaborative tensor modeling module and a marker discrimination module. According to the method, three-dimensional immune data are processed through kernel density estimation, spatial information entropy is calculated, multi-source heterogeneous medical data spatial features are integrated, a gradient map is generated in combination with a moving average method to capture a marker concentration change trend, and a boundary topology model is constructed based on directional derivative detection and morphological closed operation to recognize morphological boundary features. Three-order tensor modeling is utilized to fuse immune subtypes, time dimensions and marker expression data, a core factor matrix is extracted through dimension adaptive tensor decomposition to improve marker correlation analysis integrity, Manhattan distance is adopted to quantify collaborative expression intensity, and the Hadamard product of a residual matrix is combined to improve the correlation analysis integrity of the marker. And the crossing from single threshold judgment to multi-dimensional cooperative judgment of quantitative analysis of the immune marker is realized.

The application discloses an intelligent experiment table multi-sensing data distributed storage method based on edge calculation and relates to the technical field of edge calculation, which comprises the following steps: generating a double-layer segment matrix based on continuous sampling data of an intelligent experiment table; obtaining a tail trace residual matrix based on the double-layer segment matrix; constructing a shift matrix based on the tail trace residual matrix and generating a tail trace adhesion-propagation matrix based on the shift matrix; calculating a tail trace cutting cost based on the tail trace adhesion-propagation matrix; calculating the fitness of each particle in a particle swarm based on the tail trace cutting cost; iteratively updating the speed and position of each particle in the particle swarm based on the fitness to determine an optimal distributed write matrix; and obtaining a distributed storage result based on the optimal distributed write matrix; and the application improves the experiment data recovery efficiency in a distributed storage environment.

The application relates to the technical field of medical data analysis, in particular to an immune marker analysis system based on machine learning, which comprises an entropy density flow change module, a boundary dynamic labeling module, a collaborative tensor modeling module and a marker discrimination module.In the application, three-dimensional immune data is processed through kernel density estimation and spatial information entropy is calculated, multi-source heterogeneous medical data spatial features are integrated, a gradient graph is generated by combining a sliding average method to capture a marker concentration change trend, a boundary topology model is constructed based on directional derivative detection and morphological closing operation to identify morphological boundary features, a three-order tensor modeling is used to fuse immune subtypes, time dimensions and marker expression data, a dimension self-adaptive tensor decomposition is used to extract a core factor matrix to improve the integrity of marker correlation analysis, Manhattan distance is used to quantize collaborative expression intensity and combine residual matrix Hadamard product, and single threshold determination is crossed to multi-dimensional collaborative discrimination in immune marker quantitative analysis.

The invention relates to a distributed radar target positioning method and device, communication equipment and a storage medium, which are applied to a distributed multiple-input multiple-output radar positioning system. The method comprises the following steps: under the condition of detecting a target, constructing a residual matrix according to a bistatic distance measurement matrix, a first distance matrix between a transmitter and the target and a second distance matrix between a receiver and the target; establishing a non-convex norm minimization problem model according to the residual matrix; and approximately solving the non-convex norm minimization problem model by using a first generalized iterative reweighting algorithm to obtain a target positioning result. According to the method, the non-convex norm with the residual matrix is modeled into the target loss function, and the corresponding generalized iterative reweighting algorithm is developed to solve the non-convex minimization problem, so that the positioning precision of the MIMO radar in low SINR and NLOS abnormal environments can be ensured.