340 results about "Factor matrix" patented technology

The invention discloses an AI model training parameter dynamic optimization system, and relates to the technical field of artificial intelligence model training optimization. According to the scheme, by monitoring gradient norms in real time and fusing a frequency weighting mechanism, dynamic gradient self-adaptive cutting is achieved, the limitation of a fixed threshold value is broken through, and the model precision is guaranteed while the batch scale is expanded by 30%; a weight matrix is innovatively decomposed into a low-rank factor matrix, the internal storage is compressed to O (n + m), a strategy perception distillation technology is synchronously combined, a reward signal is dynamically generated by utilizing comparative learning to replace manual preference labeling, and collaborative optimization of parameter lightweight and knowledge migration is realized; aiming at a heterogeneous equipment environment, designing a computing power perception parameter group automatic division mechanism, and reducing communication redundancy by 40% by adopting asynchronous weighted aggregation; and constructing a data-parameter joint adjustment and optimization closed loop, and integrating a real-time data cleaning framework and a parameter normalization module to dynamically adjust the hyperparameters of the optimizer. According to the system, an efficient solution is provided for edge calculation and large model training by using a full-link adaptive architecture.

The invention relates to the technical field of equipment health state prediction, in particular to a multi-source sensing driven equipment health prediction method and system. The method comprises the following steps: synchronously acquiring equipment temperature, vibration, current and acoustic data through a multi-source sensor, carrying out denoising and standardization processing, dynamically distributing each signal weight to adapt to an equipment operation stage, generating a high-dimensional dynamic feature vector, and embedding a historical smoothing mechanism to realize continuous updating; performing standardization and nonlinear mapping on the features, constructing a dynamic coupling factor matrix to quantify a cooperative relationship between the features, fusing interaction information and adaptively enhancing abnormal features; three-layer progressive health prediction from a local part, a middle-layer subsystem to global equipment is implemented based on coupling characteristics, a trend consistency verification mechanism is introduced, global and middle-layer prediction differences are quantified through residual errors, weights are adaptively corrected, and the equipment health state evolution trend and the risk level are output. According to the method, the multi-working-condition adaptability, the feature coupling sensitivity and the prediction result reliability are remarkably improved.

The invention relates to the technical field of electric power engineering purchase demand prediction, in particular to an electric power engineering purchase demand prediction system based on machine learning, and the system comprises the steps: obtaining historical purchase data, construction progress information and electric power engineering design parameters, carrying out the standard stage division and time alignment, and constructing a stage sequence model reflecting the material use rhythm; and a coupling factor matrix is generated based on the material co-occurrence frequency and the stage position relationship, and the modeling capability of the model for the material cooperation relationship is enhanced. And the stage time sequence features, the coupling information and the structured engineering parameter vectors are fused and input into a regression prediction model, so that accurate mapping of material demands and multi-dimensional engineering features is realized, and the purchase prediction precision in a target period is improved. A deviation sequence is constructed based on historical prediction errors, and error correction is performed through a feedforward neural network, so that prediction accuracy and response capability are effectively improved, and resource waste and construction delay are reduced.

The invention discloses a tensor core, a processor, a data processing method, electronic equipment and a storage medium, and is applied to the field of tensor processing. The tensor kernel comprises a first dot multiplication unit and a scaling factor matrix multiplication processing module, the scaling factor matrix multiplication processing module comprises a second dot multiplication unit, and the first dot multiplication unit and the second dot multiplication unit support dot multiplication operations of different floating-point number precisions. The tensor core is configured to receive a first tensor, a second tensor, a scaling factor of the first tensor, a scaling factor of the second tensor, an offset term of the first tensor, and an offset term of the second tensor, perform a matrix multiplication operation using the scaling factor and the offset term using a first dot multiplication unit and a scaling factor matrix multiplication processing module, and obtaining a matrix multiplication operation result of the first tensor and the second tensor. Matrix multiplication operation using scaling factors is executed by multiplexing dot multiplication units with different precisions in a tensor kernel, extra hardware area cost is reduced, and existing hardware resources are fully multiplexed.

The invention relates to the technical field of electric digital data processing, and discloses an ALS-based parallel tensor filling method and system, and the method comprises the steps: collecting feature information of a data set needing to be filled, building a tensor according to a preset data feature, and generating an initial factor matrix of the tensor; obtaining the number and position of tensor effective values, dividing sub-tensors according to hardware equipment performance to obtain a plurality of sub-tensors processed in parallel, and storing the sub-tensors in a COO format; a processing unit is distributed to one sub-tensor, reduction operation is carried out, and updated values of each row are assigned to the currently updated factor matrix; the three factor matrixes are alternately updated and iterated for multiple rounds, after each round of iteration is completed, the updated factor matrixes are subjected to outer product to obtain filled tensors, the loss between the front tensor and the rear tensor is calculated, the factor matrixes are alternately optimized, and missing value filling is achieved; the parallel processing efficiency is improved, and the overall calculation delay is reduced.

The invention discloses an intelligent regulation and control method for a power battery positive electrode binder production process, and the method comprises the steps: synchronously collecting multi-dimensional process parameters such as temperature, viscosity and the like and performance indexes such as particle size distribution, bonding strength and the like, carrying out the normalization, denoising and time sequence alignment processing, and fusing hydrodynamic simulation and historical data to construct an initial parameter coupling model; identifying a parameter influence weight through an attention mechanism neural network, generating a decoupling factor matrix to reconstruct a parameter space, and establishing a virtual control channel; executing constrained gradient descent multi-objective optimization in the channel, generating a regulation and control instruction, and reversely mapping the regulation and control instruction into an equipment executable parameter; according to the method, performance indexes and optimization targets after regulation and control are continuously compared, model parameters are updated, retraining is triggered when deviation exceeds a threshold value, regulation and control precision and process stability are guaranteed, and the real-time performance of parameter regulation and control, the collaborative optimization capability and the quality consistency of the production process are remarkably improved.

The invention discloses a data quality treatment method and system based on an AI Agent, and belongs to the technical field of artificial intelligence and data treatment. An initial data semantic distribution map is constructed, cross-dimension correlation feature factors are extracted, a multi-scale quality anomaly sensitive factor matrix is constructed, time sequence evolution weights are embedded, and a dynamic feature evolution trajectory is formed; performing perception modeling on the evolution trajectory by using an AI Agent, generating a multi-level quality risk thermodynamic diagram, extracting a deviation dense region and constructing an anomaly propagation path set; in combination with upstream and downstream data links and task flow information, calculating a potential impact factor weight, constructing a causal traceability map, and injecting a correction strategy label for a key field node; the AI Agent autonomously selects an adaptive strategy combination according to the target data segment, and performs online intervention on the target data segment; according to the method, closed-loop treatment of the data quality problem from perception and judgment to intervention and feedback is realized, and the method has the advantages of self-adaption, high interpretability and the like.

The invention discloses a multi-drive cooperative control method for high-precision electric drive assembly equipment, relates to the technical field of intelligent manufacturing control, and is used for solving the problem of insufficient cooperative control precision of a multi-axis system under parameter mismatch and disturbance. By constructing a multi-source data fusion and dynamic coupling analysis mechanism, the system collects the motion state and control parameter data of multiple driving shafts, generates a multi-shaft collaborative dynamic response sequence, and calculates the dynamic coupling degree between the shafts to identify the uncertainty mode of the system; positioning a performance degradation shaft section by combining the temperature field and the vibration signal characteristics, and generating a system stability evaluation coefficient; and establishing a space mapping relation to construct a dynamic compensation factor matrix, and generating a control parameter adjusting quantity meeting a collaborative matching condition through fuzzy rule reasoning. According to the method, control instability caused by traditional global adjustment is avoided, precise cooperative control of the multi-axis system is achieved, and the control precision, operation stability and reliability of equipment under the dynamic working condition are remarkably improved.

The invention provides a method, a device and equipment for determining a sludge resource utilization path, belongs to the technical field of computer information processing, and solves the problem that economic, environmental and social targets are difficult to dynamically and collaboratively optimize in the current sludge resource treatment process. The method comprises the following steps: acquiring original data, wherein the original data comprises at least one of sludge attribute data, process parameters and external dynamic data; preprocessing the original data to obtain sludge sample data; classifying and integrating the sludge sample data to obtain a multi-dimensional tensor; decomposing and dynamically updating the multi-dimensional tensor to obtain a decomposed factor matrix; inputting the decomposed factor matrix into a path optimization model for processing to obtain a resource optimization path set; and performing optimal verification processing on the resource optimization path set to obtain optimal utilization path data. According to the scheme, dynamic optimal balance of sludge treatment economic benefits, carbon emission reduction and social compliance is realized.

The invention relates to the field of optical material defect detection, and discloses a PC lens defect classification and identification method and system, and the method comprises the following steps: asynchronously collecting lens multi-modal data through polarized light imaging, thermal imaging and an acoustic emission sensor, and constructing a five-order asymmetric tensor containing polarization angle, spectrum, time, space and defect features; performing dynamic dimension reduction on the tensor by using an improved Tucker decomposition method, and extracting a core tensor and a factor matrix; establishing a polarization angle-time correlation coding model based on inverse problem solution of a light field state equation, deducing defect distribution characteristics and optimizing a core tensor; mapping the optimized tensor to the QUBO Hamiltonian of a quantum annealing machine, and carrying out optimization solution; and finally, fusing the multi-mode confidence coefficients of polarized light, thermal imaging and acoustic emission signals, generating defect classification labels and inverting geometric and mechanical parameters of the defects. The lens flaw detection efficiency and the classification accuracy are remarkably improved, and the method is suitable for industrial-grade high-precision quality control scenes.

The invention relates to the technical field of spraying control, and discloses a water turbine spraying process parameter real-time monitoring, regulating and controlling method and system.The water turbine spraying process parameter real-time monitoring, regulating and controlling method comprises the following steps that multi-modal data in the spraying process is collected, and a four-dimensional data structure is constructed; tensor decomposition is carried out, and a core feature tensor and a factor matrix are extracted; constructing an enhanced game strategy model, and outputting a process parameter control strategy; lyapunov optimization is introduced, and regulation and control input is generated; the control strategy and the control input are combined to generate a spraying execution action; and performing feedforward compensation based on the disturbance observer to form a final control instruction. According to the method, the multi-modal process data fusion and tensor decomposition technology is adopted, the four-dimensional data tensor is constructed, the core features of the spraying process are extracted, and by means of the method, the system can more comprehensively capture dynamic changes in the spraying process, and an accurate technological parameter control strategy is provided.

The invention relates to the technical field of bridge engineering detection and repair, in particular to a bridge crack repair effect prediction method and system based on performance simulation, and the method comprises the following steps: obtaining basic data and environmental load data of a target bridge, and matching at least two candidate repair schemes from a preset repair scheme library; constructing an initial performance model before repair, and calculating an initial bearing capacity loss rate and a crack propagation rate; quantifying the interaction influence of the multi-modal data through a coupling factor matrix, and iteratively calculating the repair condition of a set time period after repair to obtain an initial simulation result; and collecting real-time monitoring data in a set time period in the repairing process and after repairing, and inputting the real-time monitoring data into a preset LSTM adaptive correction model to obtain a corrected simulation result. According to the method, the long-term repair condition after repair can be accurately calculated, the accuracy of a prediction result and the environmental adaptability are remarkably improved, and a scientific basis is provided for long-term service performance pre-judgment.

The invention provides a wind tunnel structure sensitive reflection settlement position intelligent identification method, and belongs to the technical field of wind tunnels. Vibration sensors and displacement sensors are arranged at key positions of a wind tunnel structure to form a monitoring network, collected signals are preprocessed, and a settlement factor matrix is established; a dynamic load matrix is constructed to describe composite load distribution, a vibration burr identification matrix is established, real signals and noise are separated by adopting wavelet transformation, a slow settlement trend matrix is constructed to extract a long-term change rule, and a least square optimization algorithm is adopted to jointly solve each matrix parameter to establish a settlement position identification function. The settlement three-dimensional position coordinate is determined according to the multi-sensor data fusion result, the structure safety state is evaluated through the settlement risk coefficient, and the technical problems that the wind tunnel structure settlement position recognition precision is insufficient, and a real settlement signal and a noise interference signal cannot be effectively distinguished are solved.

The invention relates to the technical field of data compression and storage, and discloses a monitoring data compression and storage method driven by a time sequence database, comprising the following steps: S1, preprocessing multi-source monitoring data, and constructing a dynamic tensor comprising a timestamp, a numerical index and a multi-dimensional label; s2, performing dynamic dimension reduction processing on the dynamic tensor to generate a core tensor and a multi-dimensional factor matrix; s3, based on the core tensor and the multi-dimensional factor matrix, determining a compression parameter, a decompression parallelism degree and an index granularity through a joint optimization model; and S4, performing hierarchical coding on residual data generated by the dynamic dimension reduction processing to generate a lightweight residual coding result. Through a dynamic tensor decomposition and incremental updating technology, low storage overhead and real-time dimension expansion capability of streaming monitoring data are realized, the problems of calculation redundancy and storage expansion caused by the fact that the streaming monitoring data cannot adapt to dynamic newly-added tags are solved, and meanwhile, frequent reconstruction cost caused by data dynamic expansion is avoided.

The invention relates to the field of intelligent operation and maintenance of offshore wind power construction ship equipment, and discloses a remote monitoring and fault diagnosis and repair system for offshore wind power construction ship equipment, and the system comprises a data collection module which is used for collecting the operation data of the ship equipment through a multi-mode sensor, and outputting the data to a federal learning module; the federated learning module is used for receiving the output data of the data acquisition module, performing tensor decomposition on the multi-modal data to obtain a local factor matrix, uploading the factor matrix to a cloud server for global aggregation, and issuing the aggregated global factor matrix to the fault diagnosis module; and the resource allocation module is used for dynamically receiving the communication load and the edge computing requirement of the federated learning module. Through a multi-mode sensor network and an anti-interference preprocessing algorithm, the influence of severe environments such as offshore high salt mist and strong electromagnetic interference is effectively overcome, and continuous and stable acquisition of equipment operation data is ensured.

The invention relates to an intelligent driving control method of a vehicle and the vehicle, and belongs to the technical field of intelligent driving, and the method comprises the following steps: obtaining an original data stream of a vehicle sensor network, and carrying out feature extraction on a dynamic feature vector; state vectors of the traffic participants are extracted according to the dynamic feature vectors, edges between the two traffic participants are constructed, a traffic participant interaction graph is constructed, and an asymmetric factor matrix is calculated; acquiring a historical scene data set of the vehicle, calculating correction similarity, and calculating a weight coefficient; performing weighted fusion on the weight coefficient and the historical scene data set to obtain an enhanced data set; and performing updating training on the pre-trained intelligent driving decision model according to the enhanced data set to obtain an updated intelligent driving decision model, generating an intelligent driving decision, and controlling vehicle operation, thereby realizing more accurate characterization of a dynamic game relationship in a complex traffic scene, and improving the accuracy of the dynamic game relationship. And the intelligent driving decision model can continuously adapt to an asymmetric interaction mode effect in a real scene.

The invention discloses an energy storage system intelligent energy management control method and system, and relates to the technical field of energy storage system intelligent energy management control, and the method comprises the steps: obtaining an energy storage unit operation state parameter and an environment disturbance factor matrix through a multi-source data collection device, and carrying out the data preprocessing; and constructing a long-short-term memory neural network model, dynamically adjusting a prediction time window of the model and optimizing a target function weight coefficient based on the load predicted by the model, and determining a charging and discharging control instruction. And updating and optimizing the target function weight coefficient and the charging and discharging strategy library through a reinforcement learning algorithm to realize self-adaptive optimization adjustment of the charging and discharging strategy. According to the method, high-precision energy management and intelligent optimization control of the energy storage system in a complex environment are realized. The load prediction accuracy and the energy utilization rate of the system are improved, the aging rate of the battery is reduced, the service life of the battery is prolonged, the adjusting capacity of the energy storage system is improved, and the overall stability of the energy storage system in dynamic change is enhanced.

The invention relates to the technical field of nondestructive testing, and discloses an aircraft defect identification method and system based on tensor decomposition and an attention mechanism, and the method comprises the following steps: obtaining multi-modal data of an aircraft; constructing the multi-modal data into a fourth-order space-time-modal tensor, and generating a dynamic graph structure based on the modal characteristics of the tensor; applying mixed constraint to obtain a core tensor and a factor matrix during four-order tensor decomposition; features are extracted through multi-scale pooling, weights are distributed in combination with topological persistent coherence and a gating attention mechanism, and feature fusion is achieved; and identifying defect types based on fusion features, and positioning defect regions by using factor matrix gradient amplitudes and dynamic thresholds. According to the method, through multi-modal data fusion, dynamic graph regularization constraint and mixed tensor decomposition technologies, the detection sensitivity and the positioning precision of the small defects on the surface of the aircraft are improved, and meanwhile, the physical interpretability of the characteristics and the robustness of the algorithm to complex working conditions are enhanced.

The invention discloses a patient data identification method based on multi-dimensional feature fusion, a medium and equipment. The method comprises the following steps: firstly, extracting identity features, time sequence features and clinical features from patient diagnosis and treatment records; then calculating a basic similarity based on the identity features and the clinical features, and screening out a first diagnosis and treatment data set which is preliminarily matched; then, a suggested review period is determined by analyzing the diagnosis and treatment scheme, a period compliance weight is generated in combination with the deviation degree of an actual review interval and the suggested period, and meanwhile, a stage continuity weight is generated according to the diagnosis and treatment stage incidence relation; carrying out weighted summation on the two weights to obtain a weight influence factor, and constructing a weight influence factor matrix through iterative calculation; and finally, adjusting the basic similarity by using the matrix to obtain a corrected similarity, screening out a finally matched second diagnosis and treatment data set, and distributing a unique identifier. According to the method, the accuracy and the reliability of patient data identification are remarkably improved by fusing multi-dimensional features and introducing a dynamic weight adjustment mechanism.

The invention relates to the technical field of positioning navigation, in particular to a vehicle-mounted IMU (inertial measurement unit) data compensation method for intelligent driving inertial navigation, which comprises the following steps: acquiring various attitude data of a vehicle-mounted IMU through a rotating platform, acquiring a reference quantity, establishing a target function, and acquiring an offline calibrated rotating error matrix and a scale factor matrix by adopting an optimization algorithm; the method comprises the following steps: acquiring operation original measurement in a vehicle operation process, defining a system state vector, analyzing a Taylor expansion approximation relation between a cross multiplication matrix corresponding to an error small angle and a rotation error matrix, and constructing a discrete time state equation and an observation equation based on an IMU kinematics model; acquiring an error small angle after each update by adopting extended Kalman filtering; and carrying out real-time compensation on the vehicle-mounted IMU data in combination with the operation original measurement at each moment. The invention aims to improve the precision and robustness of inertial navigation and meet the requirement of intelligent driving for continuous and stable positioning.

The invention relates to the technical field of wireless communication, and discloses a wireless communication adaptive method and system based on a data transmission state, and the method comprises the steps: obtaining multi-source data, carrying out the preprocessing of the multi-source data, obtaining a CSI compression matrix, predicting the channel coherence time, dynamically adjusting the CSI sampling interval, and defining a load-channel coupling factor; obtaining cross-layer data based on the CSI compression matrix, performing fusion through a rotation matrix to obtain a fusion matrix, and extracting a physical layer fusion feature and an application layer fusion feature to calculate a multi-target state score; establishing a 5G power compensation mechanism based on cross-layer data, calculating a four-dimensional influence tensor, performing tensor decomposition and optimal action selection, and decomposing T into a core tensor and a factor matrix; selecting an optimal parameter combination through modular product calculation; based on cross-layer data, a quantum entanglement feedback mechanism is introduced, data is fed back, entanglement state association cross-layer indexes are designed, a quantum gate is adjusted through entanglement state design, and a model is updated in real time in combination with incremental learning.

The invention relates to the technical field of computer data processing, and discloses a building material supplier dynamic evaluation and recommendation system based on big data, and the system comprises a data fusion module which integrates multi-source heterogeneous data to generate a unified data set; the tensor modeling module is used for constructing and decomposing a five-dimensional space-time tensor to obtain a factor matrix and a dynamic weight; the causal correction module is used for establishing a causal graph based on the network relationship and eliminating hybrid deviation; the recommendation decision module is used for outputting a recommendation list through reinforcement learning in combination with the evaluation weight and the performance distribution; and the interpretable module is used for generating an interpretable report based on the factor and the causal path. According to the method, the technical scheme of multi-source heterogeneous data fusion and five-dimensional space-time tensor decomposition is adopted, and dynamic, multi-dimensional and relevance evaluation of supplier performance is realized by constructing a unified data structure including suppliers, time, static characteristics, context and cooperative relationships.

The invention relates to a quick mountain torrent disaster risk assessment method and system based on multi-source data fusion. The method comprises the following steps: performing space-time registration on multi-source data, constructing a dynamic factor matrix and a static geological parameter matrix, and calculating soil saturation; on the basis of the dynamic factor matrix, removing redundant factors, analyzing and generating a weight vector through an entropy method, on the basis of the effective factor set, the static geological parameter matrix and the soil saturation, correcting the rock-soil shear strength by quantifying the sediment deposition effect, obtaining the corrected rock-soil shear strength, and calculating the slope safety coefficient; and according to the coefficient, the effective factor set and the weight vector, generating a risk membership degree vector through fuzzy evaluation, calculating a risk feature value, identifying a disaster dominant type, and constructing a risk map. According to the method, through multi-source data space-time registration, redundant factor elimination, sediment deposition effect quantitative correction and the like, the efficiency and precision of mountain torrent disaster risk assessment are improved, and the support of an assessment result on disaster prevention and reduction decisions is enhanced.

The invention discloses a multi-dimensional layered current limiting method based on spatio-temporal feature fusion. The method comprises the following steps: acquiring a hardware limit speed, and generating a global quota based on a reinforcement learning model; calculating a variable coefficient, and if the variable coefficient exceeds a preset critical value, introducing a conservative coefficient and generating a weight factor through nonlinear function mapping to suppress the micro network jitter; constructing a priority factor matrix based on the service priority and the service type, and obtaining a service priority factor; correcting the average bandwidth of the node by integrating the multi-dimensional dynamic weight factor and the service priority factor to obtain a suggested speed, inputting a historical performance index into the LSTM model to predict the load of the next period, and if the historical performance index exceeds a critical value, triggering connection migration before executing current limiting to realize active avoidance; otherwise, comparing and selecting the minimum value of the global quota, the suggested speed and the hardware limit speed as the final current-limiting speed. The problems of current limiting strategy lag and wide oscillation can be solved, and the bandwidth utilization rate and the system stability are improved.

The invention relates to the technical field of power grid dispatching, in particular to a power grid dispatching operation ticket intelligent generation and risk pre-judgment system based on multi-modal data fusion. The method comprises the following steps: a weather-time window mapping unit associates and identifies weather data on that day, and judges the rationality of planned operation time and steps by establishing a mapping relationship between a meteorological condition and an operation time window and utilizing a meteorological influence factor matrix; after the historical operation ticket analysis unit confirms meteorological condition operation, information is extracted from the meteorological condition operation through the NLP technology, and a set of standard operation ticket template library is constructed; the four-dimensional risk association unit establishes a four-dimensional risk association model based on power safety regulations, historical accident cases and equipment defect data. According to the design of the invention, through multi-dimensional data fusion technologies such as meteorological-time window mapping, four-dimensional risk association analysis and dynamic topology simulation, safety risks existing in the operation process can be comprehensively identified and evaluated.

Disclosed in the present application are a compression and training method and apparatus for a defect detection model. The method comprises: obtaining, by means of segmentation labeling, a segmentation labeling factor matrix of each sample image; inputting each sample image into both a first defect detection model and a second defect detection model, and extracting first feature maps outputted by target convolutional layers in the first defect detection model and second feature maps outputted by corresponding target convolutional layers in the second defect detection model; and calculating, by using the segmentation labeling factor matrix, corrected distances between corresponding feature vectors of the first feature maps and the second feature maps, and calculating, as a first loss function, the sum of the corrected distances between all the feature vectors of the first feature maps and the second feature maps. The present embodiment can improve the accuracy of detecting tiny product appearance defects by means of a compressed defect detection model.

A hyperspectral image change detection method based on spectral grouping tensor decomposition reconstruction comprises the following steps: splicing a pair of preprocessed dual-time hyperspectral images in a spatial dimension, extracting histogram features on all spectral channels, determining the number of groups, and grouping the spectral channels according to the histogram features; performing Tucker decomposition on each group of spectral channels of each hyperspectral image to obtain a core tensor and a factor matrix, performing principal component truncation on the decomposed core tensor and factor matrix according to an accumulated singular value ratio, and reconstructing the image; and after the reconstructed image is sent to a change detector to obtain a distance map, a threshold method or a clustering method is adopted to process the distance map, and a final change map is generated and output. According to the method, spectral correlation priori knowledge is fully utilized, low-rank information can be better extracted from grouped data through Tucker decomposition and reconstruction, a more accurate and more robust change detection result can be obtained, and the influence of factors such as image registration errors and noise on the detection performance is reduced.

The invention discloses a gravitational field inversion method considering unsteady noise of a satellite gravity observation value. The gravitational field inversion method comprises the following steps: A, acquiring original observation data of a gravity satellite; b, setting observation value noise as white noise, and constructing a random model considering non-stationary noise; c, performing parameter estimation to obtain a post-test residual error of an observation value; d, estimating the noise auto-covariance and precision factor matrix of the observation value based on the post-test residual error of the observation value; e, repeating the steps C and D for iteration to realize refinement of the gravitational field parameter and observation value random model; and F, evaluating the time-varying gravity field model in the spectral domain. According to the invention, the defects in the prior art can be improved, and the precision and stability of gravitational field inversion are improved.

The present invention discloses a deformation prediction method for shield construction passing through buildings based on multiple factors. A multi-factor matrix is ​​established by using settlement data, tilt data, the distance between the construction face and the building, shield construction parameter data, and geological information data of the construction area obtained at multiple times. The matrix is ​​used as the input of a prediction model to obtain the predicted value of the settlement data of the building, thereby realizing deformation prediction of shield construction passing through buildings based on multiple factors. The present invention combines CNN and LSTM models to not only correlate the characteristics of multiple factors, but also correlate the progress of shield construction with the deformation of the building through the factor of the distance between the shield construction and the building. These technologies can not only effectively reduce construction risks, but also improve overall construction efficiency and economic benefits, making settlement prediction more intelligent, dynamic, and refined. It can greatly improve the prediction accuracy and safety of the settlement of adjacent buildings during shield construction.

The invention discloses an energy storage system fault prediction method and system based on multi-modal data fusion, and relates to the technical field of energy storage systems. Comprising the following steps: acquiring multi-modal data, and preprocessing the data to improve the data quality; performing feature extraction on the multi-modal data through a feature extraction model to obtain a feature matrix; performing factorization on the extracted feature matrix, establishing a multi-modal data fusion model, fusing core factor matrixes obtained after decomposition, establishing a prediction model, substituting fused feature vectors for fault prediction, and outputting a prediction result. According to the method, factorization is carried out on the extracted feature matrix through a data decomposition method, the feature decomposition matrix is obtained, core factor matrixes obtained after decomposition are fused through the multi-modal data fusion model, the fused feature vector can further express the mutual relation between multi-modal numerical values, and the fusion efficiency is improved. And the fault prediction accuracy of the energy storage system is improved.