289 results about "Large sample" patented technology

The invention relates to a circuit board on-line defect detection method and system, and the method comprises the steps: carrying out the synchronous collection and structural integration of multi-source technological parameters such as production line environment temperature and humidity, equipment operation states, material batches and the like, and defect detection images, and achieving the construction of large-sample original data in a production process; through standardization and de-noising preprocessing, multi-modal features are fused, and a distribution mapping model of process and defect features is established by using algorithms such as mutual information analysis and principal component analysis. Based on a feature distribution model and real-time data, a dynamic anomaly detection threshold is adaptively generated, Bayesian inference and evidence reasoning are combined, multi-level confidence levels and risk response suggestions are output, and self-learning evolution of the model and the threshold is realized through a closed-loop feedback mechanism. According to the scheme, the accuracy of anomaly detection, the response timeliness and the risk disposal intelligent level are improved, the method adapts to complex working conditions and batch changes, and the closed-loop optimization and safety control capability of the production process is remarkably enhanced.

The invention provides a bearing fault variable working condition diagnosis method based on a multi-scale convolutional network and MAML, and relates to the technical field of equipment fault diagnosis. The method comprises the following steps: firstly, introducing fast Fourier transform to pre-process an original time domain vibration signal; secondly, a fault diagnosis model based on a multi-scale convolutional network and MAML is applied; and then, an internal and external circulation updating method based on model-independent element learning is adopted, so that the model can quickly adapt to a new task, and a relatively good prediction effect can be achieved only through a small amount of fine adjustment. According to the method, multi-scale feature extraction and meta-learning are creatively combined, the generalization ability of the model under variable working conditions is remarkably improved, the problem that a traditional fault diagnosis method depends on a single working condition and a large sample size is effectively solved, and the method is particularly suitable for small-sample and multi-working-condition bearing fault diagnosis scenes on an industrial site.

The invention relates to the technical field of medical informatics and artificial intelligence technology cross application, in particular to a dynamic intelligent prediction system for HIV infected person immune reconstruction insufficiency. Comprising a data acquisition and management module, a data preprocessing and feature engineering module, a Bayesian joint modeling core module, a dynamic prediction calculation module, a multi-dimensional verification and evaluation module and a clinical application service module. According to the method, on the basis of large-sample multi-center longitudinal follow-up visit data, a Bayesian shared parameter dynamic joint model is adopted, the dependency relationship between dynamic trajectories of immune indexes such as cell counting and the like and IIR occurrence time is captured, real-time risk prediction is achieved through a sequential Bayesian updating mechanism, and through multi-dimensional verification, the risk prediction accuracy is improved. The method has the advantages that the performance is better than that of an expert-driven model, expert experience pre-judgment and 19 mainstream machine learning algorithms, an individualized conclusion with a confidence interval can be output, clinical decision is assisted, prediction is promoted to be clinical from scientific research, and the method has important application value and popularization prospect.

The invention discloses a method for acquiring grain orientation characteristics based on optical and height characteristics of an alloy etching structure. The method comprises the following steps: (1) acquiring grain orientation distribution and grain size of an alloy surface structure; (2) obtaining an etching structure on the surface of the alloy; (3) acquiring optical characteristics of the etched structure on the surface of the alloy; (4) obtaining height characteristics of an alloy surface etching structure; and (5) constructing a correlation model of structure grain orientation characteristics based on alloy surface optics and height characteristics, and realizing prediction. Aiming at the test of the alloy grain orientation, the optical and height characteristics of the alloy surface etching structure are synthesized, the measurement of the centimeter-level large sample size or full-area grain orientation of the alloy is easy to realize, and compared with the existing measurement aspect, the method is more efficient and convenient; and more comprehensive and reliable tissue information can be provided for alloy optimization and new material development, and the method has important significance for high-throughput characterization and intelligent manufacturing.

A data processing method, applied to the field of artificial intelligence. The method comprises: acquiring data and a corresponding label, the label corresponding to an expert rule; and under the condition that the number of data sets corresponding to the label is smaller than a first threshold value, inputting into an algorithm model prompt information composed of the data sets, the label, and the expert rule to generate new data corresponding to the label, wherein the prompt information comprises the label, the data, and the expert rule; and there is a mapping relationship among the label, the data, and the expert rule. According to the present application, the prompt information prompt comprising the expert rule is used as the input of the algorithm model, the algorithm model can be guided to generate new data so as to expand small sample data, and the expanded small sample data set is fused with a preprocessed large sample data set, realizing the balance of data sets of different labels. Meanwhile, under the condition that the diversity and balance of the data are ensured, the performance of the model is improved by combining the dynamic optimization of the prompt information prompt.

The invention provides an engineering drawing steel bar bulk sample annotation text detection method, system and device based on deep learning, and a storage medium. The method comprises the following steps: S1, preprocessing an engineering drawing image and labeling a steel bar bulk sample annotation text region data set; s2, constructing a staged multi-scale feature map extraction backbone network framework; s3, introducing compression excitation into each scale feature map; s4, adopting an ACON adaptive activation function in nonlinear modeling; s5, constructing a multi-scale feature pyramid structure based on the RSE-FPN, and superposing scales to reinforce fusion; and S6, outputting the approximate binary image to realize accurate prediction of the bounding box. According to the method, the OCR detection precision is remarkably improved, small character and complex background interference is effectively overcome, and missing detection and false detection are reduced; the model is lightweight to facilitate efficient deployment of equipment; the output boundary is compact and accurate, the semantics is reasonable, and subsequent recognition and analysis are facilitated; the method has excellent generalization ability, adapts to steel bar large sample drawings of different formats, definitions and styles, and is high in engineering practicability.

The invention provides a celluloid nitrocotton stability quantitative determination method based on conductivity analysis, and the method comprises the following steps: determining the conductivity of standard nitrogen oxide gas dissolved in an aqueous solution by adopting a conductivity meter through a standard gas simulation mode according to the criterion of Beckmann-Lockg method 2mL / g; taking the numerical value as a stability index critical value of the celluloid nitrocotton; the method comprises the following steps: absorbing nitrocotton with the same mass by using ultrapure water, heating at 132 DEG C for 2 hours, releasing nitrogen oxide gas, recording the change value of the conductivity in the solution along with time in real time by using an oxidation stability tester, obtaining the change value of the conductivity after 2 hours, and determining the content of nitrogen oxide according to the comparison between the conductivity value and a critical value. And judging whether the stability of the celluloid nitrocotton sample is qualified or not. The method provided by the invention can solve the technical problem that accurate quantitative judgment is difficult to realize due to large sample quantity, high safety risk, tedious operation process and large color change error of test paper observed by human eyes in the traditional method.

A method and apparatus for scanning probe-based surface characterization is disclosed which allows high scanning velocities and data throughput and is suitable for large samples. A combination of rotational and linear translation is used for scanning the probe by concentric circle or spiral trajectories in overlapping ring patterns. A method and apparatus for rapid control of tilt-related probe-sample separation distance and suitable for such scanning pattern is described.

The invention relates to the technical field of deep learning, and particularly provides a whale optimization-based power grid fault positioning method and system, a terminal and a medium, and the method comprises the steps: obtaining detection data of a plurality of power devices in a first time period, the detection data comprising feature values of detection signals of a plurality of sensors; performing denoising and dimension reduction processing on the detection data to obtain sample data; training a pre-constructed convolutional neural network based on the sample data by using an improved whale optimization method to obtain a fault positioning model; wherein the fault positioning model is used for determining the fault position of the power grid according to the detection data of the multiple pieces of power equipment within the second time period; the first time period is before the second time period. According to the method, the adaptability of the convolutional neural network in processing large sample data is improved, so that accurate positioning of a power grid fault is realized.

The invention provides a data stream active sampling and predicting method for a product ash content index in a coal dense medium separation process. The method comprises the following steps: S1, preprocessing coal dense medium separation data; s2, establishing a dynamic cache window in the sorted data stream to perform synchronous clustering; s3, establishing a feature memory dictionary, and constructing a centralized kernel space by using data in the feature memory dictionary to obtain a dictionary data feature micro-mode direction and a feature main mode direction; s4, calculating the projection of the real-time data point in the feature micro-mode direction, and judging whether active sampling is carried out or not; s5, obtaining the projection of the dictionary data in the main mode direction by using the kernel principal component information in the S4, constructing an evaluation function, and judging whether to sort active sampling or not; s6, taking the maximum sample set of the value function as a new feature memory dictionary; and S7, strengthening the weight of the actively sampled sample, recursively constructing a random weight neural network through a Woodbury formula, and predicting the clean coal ash content in the label data stream until the data stream is finished.

The invention provides a gas sensor response fingerprint feature extraction method based on parameter fitting, and relates to the technical field of gas fingerprint feature extraction. According to the method, curve parameter fitting is carried out on response data of a gas sensor to obtain feature sequences, the information entropy contribution rate of the features is calculated based on the inter-class and intra-class divergence ratio of each feature, the feature sequences are sorted and screened according to the information entropy contribution rate, and therefore high-compactness and interpretable fingerprint features are obtained. The feature extraction method and the extracted fingerprint features have interpretability, and the method has lower calculation and development cost and is suitable for feature extraction of a large sample set and feature extraction of a small sample set.

The small sample anti-migration prediction method suitable for the metallurgical process end point component under the zero expansion characteristic comprises the steps that smelting report data of a large sample steel grade and a small sample steel grade in the metallurgical process are collected to serve as source domain data and target domain data; adopting median to fill and restore abnormal values existing in the report data; dividing the source domain data and the target domain data into continuous feature variables and classification feature variables; inputting the continuous feature data and the classification feature data of the source domain data into a TabNet coding and decoding self-supervising network for self-supervising training, and inputting the target domain data into the trained self-supervising network for feature extraction and reconstruction; introducing an adversarial network, and gradually aligning feature distribution of a source domain and a target domain; and inputting the high-dimensional reconstruction features processed by the TabNet coding and decoding self-supervised network in the source domain into the deep table network model for training, and inputting the small sample steel grade features aligned by the adversarial network distribution into the trained deep table network model for transfer learning.

This invention discloses a combination of SNP loci, a gene chip, and their applications for cotton variety identification, relating to the fields of bioinformatics and molecular breeding. From 100 cotton whole-genome resequencing data, this invention screened out 128 SNPs, which can be used as SNP fingerprints for cotton varieties. Using the SNP molecular markers and liquid-phase probes developed in this invention, cotton identification achieved a 100% detection rate, with a minimum allele frequency greater than 0.05, capable of distinguishing any two cotton varieties, enabling efficient and rapid cotton variety identification. Furthermore, this invention utilizes next-generation sequencing technology, resulting in a simple and standardized procedure that reduces human error, leading to extremely low unit costs for batch identification analysis. It is economical and efficient, particularly suitable for large-sample analysis and identification, and has promising application prospects.

The invention relates to the field of signal processing, and discloses a noise reduction method and system based on millimeter wave radar signals, and the method comprises the following steps: collecting millimeter wave radar signals to be denoised; the method comprises the following steps: constructing a fitness function of a ratio of a mean value to a variance of a multi-scale Kolmogorov entropy based on a millimeter wave radar signal; optimizing a decomposition mode number K and a penalty factor alpha of variational mode decomposition by using a starfish optimization search algorithm; variational mode decomposition is carried out on the millimeter wave radar signal; the sample entropy of each intrinsic mode component is calculated, the mode position with the maximum sample entropy break variable is determined, the intrinsic mode components with the sample entropy smaller than the position serve as signal modes, and the rest serve as noise modes; carrying out wavelet threshold denoising on the noise mode; and performing signal reconstruction by using the signal mode and the de-noised noise mode, and outputting a de-noised radar signal. The millimeter-wave radar gas leakage signal noise reduction method solves the problem that the noise reduction effect of millimeter-wave radar gas leakage signals is poor in the prior art, and has the advantage of being capable of improving the signal-to-noise ratio of the signals.

The invention provides a crack pouring glue compression rebound recovery rate test system and method, and relates to the technical field of rebound recovery rate test.The crack pouring glue compression rebound recovery rate test system comprises a tester, a high-precision injection ball and a large sample containing vessel; the tester comprises a machine body, a touch control type control display screen is mounted on an inclined panel of the machine body, a group of function buttons are mounted on the inclined panel of the machine body, the machine body is provided with a stepped circular groove, and the large sample containing vessel is matched with an upper side circular groove of the stepped circular groove; the machine body is connected with a U-shaped frame, and the U-shaped frame is connected with a first laser distance measuring sensor. Aiming at the defects in the prior art, the invention develops the system and the method for testing the compression resilience recovery rate of the crack pouring glue, and the high-precision penetration ball is adopted, so that the contact area with a sample is increased, and the detection error is reduced. By optimizing a hardware structure (replacing a high-precision sensing assembly) and control logic, automation and high precision of a test process are realized.

The invention relates to the technical field of bridge structure health monitoring and damage diagnosis, and discloses a bridge damage identification method based on a physical information neural network. According to the bridge damage identification method based on the physical information neural network, efficient and accurate identification of bridge damage is realized by optimizing a data acquisition scheme, designing a bridge exclusive physical constraint, constructing a biterm loss function and standardizing a network training process; the invention provides a damage identification method fused with a physical information neural network, aiming at solving the problems of strong manual dependence, large sample labeling demand, weak anti-interference capability and low identification precision in the existing bridge damage identification technology. The bridge structural mechanics principle and the neural network technology are deeply combined, so that the accuracy, robustness and engineering practicability of damage identification are improved while the dependence of the labeled sample is reduced, and the actual demand of real-time health monitoring in the bridge operation stage is met.

The invention discloses a method for automatically collecting nucleic acid of aquatic microorganisms. According to the invention, a water sample is injected into a coagulation reaction bin with a temperature control function, and microbial cell lysis is realized through intermittent low-speed stirring; injecting a surface modification magnetic bead suspension into the reaction bin, starting a multi-shaft stirrer to realize efficient combination of a nucleic acid-magnetic bead compound, and synchronously starting temperature control in the bin to maintain RNA stability; by integrating high-precision positioning, pressure sensing and temperature control cracking core technologies, whole-process standardized operation is realized. The device adopts a dual-mode positioning and pressure sensor to ensure millimeter-level precision of geohydrological data of a sampling point; the three-dimensional turbulent flow field is combined with the magnetic separation array, so that the nucleic acid capture rate is greatly improved, and the risk of nucleic acid degradation is reduced as much as possible. According to the modularized flow path system, through three-stage gradient washing and nanometer ultraviolet sterilization, the problems that in traditional water sample collection, the sample size is large, information is prone to being lost, and treatment lags are effectively solved, and the overall collection efficiency and collection accuracy are improved.

The invention provides a turbine blade fatigue performance influence factor analysis method and system and a medium, and belongs to the technical field of turbine blade performance prediction.The turbine blade fatigue performance influence factor analysis method includes the steps that a small number of training samples are generated, and a back propagation neural network (BP neural network) describing the relation between the fatigue life of a turbine blade and input variables is constructed; a training sample is added to the sequence, and the BP neural network is updated until the calculated failure probability is converged; based on a failure sample obtained in the process of solving the failure probability, estimating conditional failure probability estimation values at different sample points by using a Bayesian inference theory; and obtaining a fatigue reliability sensitivity estimation value of each input variable according to an average difference between the fatigue failure probability estimation value of the turbine blade and the condition failure probability estimation value of each failure sample point. The method solves the problems that when an existing agent model method is used for solving the reliability and sensitivity of the turbine blade, the sample requirement is large, efficiency is low, and the method cannot be suitable for a high-dimensional problem.

The invention discloses a large-scale composite member deformation gap distributed measurement and full-field reconstruction method, and the method comprises the following steps: setting a plurality of sets of technological parameters, and carrying out the mold pasting of a composite member based on the technological parameters, so as to enable the surface of the composite member to form a skin; determining an optimal arrangement scheme of the sensors, arranging the sensors according to the scheme, and obtaining a gap value matrix; obtaining a plurality of groups of gap value matrixes under different process parameters, and constructing a finite element model small sample training set; training the finite element model by using the finite element model small sample training set to obtain a full-field gap prediction finite element model; and generating a convolutional neural network large sample training set through the full-field gap prediction finite element model, and training the convolutional neural network by using the large sample training set to obtain a full-field gap prediction convolutional neural network model. The method has high systematicness and engineering implementability, and is more suitable for film pasting quality evaluation of a complex shell surface structure compared with a traditional method.

The invention provides a method for diagnosing early lung cancer based on a terahertz sensor array and an artificial intelligence technology. 16 terahertz sensors with the same structure and high-quality factors (Q value) are adopted to form a 4 * 4 array, and the 16 lung cancer markers are specifically detected through modification by a molecular imprinting technology. During detection, the modified terahertz sensor array is placed in a closed gas chamber made of polytetrafluoroethylene, nitrogen is firstly introduced to obtain a reference spectrum, then gas exhaled by a subject is introduced to obtain a test spectrum, and the concentration of the marker is reflected through the harmonic peak frequency shift amount. An initial diagnosis model is established in combination with individual information of a subject, and the model is trained and optimized by using an artificial intelligence technology and a large sample. According to the method, the problems of insufficient specificity, low sensitivity, high misdiagnosis rate of a single marker and the like in a traditional method are solved, and non-invasive, rapid and accurate diagnosis of the early lung cancer is realized.

The utility model relates to a portable animal sample storage transfer box which comprises a box body, a refrigeration cavity is arranged between the inner cavity and the outer wall of the box body, the inner cavity is divided into partition cavities in an array shape through partitions which are arranged in a transversely and longitudinally staggered mode, sample boxes are inserted into the partition cavities, inner sliding grooves are vertically formed in the side walls of the sample boxes, and the inner sliding grooves are communicated with the partition cavities. A sliding block is slidably connected into the inner sliding groove, the outer end of the sliding block protrudes out of the outer wall of the sample box and is hinged to a marking belt, a spring is fixedly arranged at the bottom end of the sliding block, the bottom end of the spring is fixedly connected with the bottom end of the inner sliding groove, and a through groove for sliding of the outer protruding part of the sliding block is formed in the outer wall of the sample box; the portable animal sample storage and transfer box is reasonable in structure, larger in sample information recording interval, richer in information recording and convenient to observe, distinguish and trace, and sampling tube concave cavities and sampling bottle concave cavities for the sampling tubes and the sampling bottles to be inserted and embedded are formed in the sample boxes.

The invention relates to a large sample exploratory experiment conclusion macro-micro interaction verification method and system, the method is used for constructing a'macro-interaction-micro 'three-layer linkage analysis framework, and the framework comprises a macro conclusion layer, an interaction analysis layer and a micro instance layer; the interaction analysis layer is used for connecting the macroscopic conclusion layer and the microscopic instance layer; the method comprises the four steps of automatic discovery and presentation of a macroscopic conclusion, interactive deep exploration and verification based on the macroscopic conclusion, macro-micro bidirectional linkage and sample screening, and concrete redisk and causal verification of micro instances. Through the method, interactive verification of a macroscopic conclusion and a microcosmic instance is realized, and the interpretability and credibility of the conclusion are remarkably improved. According to the method, a macroscopic and microcosmic barrier is broken through, a cognitive closed loop from'knowing the course 'to'knowing the course' is constructed, and a more robust basis is provided for decision making.

Systems, methods, and tools associated with Enhanced Object Mask (SEI) messages are disclosed herein. In an example, a video device may obtain one or more syntax elements associated with video data. The one or more syntax elements may specify a parameter (e.g., a location parameter) associated with an object mask associated with the picture. The parameter may correlate with at least a sample range of object mask samples around the object mask ID value. In an example, the parameters associated with the object mask may include a maximum sample value and a minimum sample value of the object mask that define a range of the object mask samples around the object mask ID value. In an example, the parameter associated with the object mask may be a bounding box parameter associated with the object mask.

This invention discloses an efficient RNA data preprocessing method to address the problems of low processing efficiency and low information density in high-throughput sequencing data. Its core steps include: (1) introducing a parallel processing scheme for high-throughput sequence data, rapidly mapping RNA-seq data to a reference genome to generate a BAM file; (2) extracting base sequences and expression levels and storing them as compact PKL format files; (3) extracting all exon position information by parsing the genome annotation file; (4) combining multi-sample expression level data to screen for highly variable exons and constructing a high-information-density feature list based on the sample set; and (5) accurately extracting target sequences from the preprocessed file based on this list. Compared to traditional methods, this innovative approach achieves triple optimization: full-process parallel processing for accelerated computation, high-compression data storage, and adaptive feature selection. Processing speed is increased by 3-5 times, and data volume is reduced by more than 90%, making it suitable for high-throughput RNA-seq data analysis with large sample sizes.

The invention particularly relates to a chip co-culture method, a high-content imaging analysis method based on the chip co-culture method and application of the chip co-culture method in anti-tumor drug screening. The tumor-related fibroblasts and the tumor organoid cells are respectively subjected to resuspension counting with a PBS buffer solution, the two kinds of cells are fully and uniformly mixed according to the quantity ratio of 1: (2-3) to form co-culture cell suspension, and the co-culture cell suspension is inoculated into a chip according to the volume dose of 1.5-2 mL per row; placing the culture chip in a 5% CO2 incubator at 37 DEG C for first culture, after incubation for 20-30 minutes, slightly and completely sucking PBS in the chip by using a pipette, adding 1.5-2 mL of an organoid culture medium into each hole of the chip, and placing the chip in the 5% CO2 incubator at 37 DEG C for second culture to obtain the co-culture cell system. Rapid and high-quality detection of a large amount of samples can be realized, and an important technical support is provided for deeply understanding a tumor microenvironment and developing a new treatment strategy.

The invention belongs to the technical field of high polymer materials, and particularly relates to a method for predicting polymer performance parameters. The influence of the sequence structure of the polymer system on the performance of the polymer system is deeply considered, and the polymer system is established by full-atom simulation; in order to save the cost, a small sample data set is obtained only through full-atom simulation; secondly, a feature representation method is provided, the relation between all monomers in the sequence structure is considered, and a WGAN-GP model is used for expanding a small sample data set so as to enlarge the sample data set; based on the expanded data set, the polymer system glass transition temperature prediction method with higher accuracy, higher efficiency and stronger generalization ability is provided. The method has the advantages of simplicity and convenience in operation, rapidness, high efficiency, high accuracy and the like, can effectively reduce time cost and experimental workload, and can be used for guiding design of high-performance materials.

The invention discloses an unbalanced data regression method and system, a storage medium and a terminal, and belongs to the technical field of data regression, and the method comprises the steps: constructing a grid division-based multi-granularity space generation model, and dividing similar samples in the same granularity space; oversampling is independently carried out in each granularity space, and an oversampling range is defined by combining the space importance degree and the dynamic neighborhood radius. According to the method, under the scene of large sample size, the data stream and imbalance problems are comprehensively considered, the processing efficiency is improved, the overall distribution of data is considered, the interference of noise samples is reduced, the continuity and internal correlation of the samples are reserved, and a new thought is provided for the field of imbalance data stream regression.

The present invention discloses a classification method and system for the invasive evolution of lung adenocarcinoma based on a state space network, which relates to a classification method for lung adenocarcinoma infiltration. The purpose is to solve the technical problems of poor classification accuracy and low classification efficiency of the IA type classification of GGO imaging features in the prior art. The constructed GGO invasiveness classification model includes a teacher encoder, a student encoder, and a fine-tuning encoder, and the teacher encoder, the student encoder, and the fine-tuning encoder are all SSM state space networks; during training, a large sample data set without GGO classification labels is used to train the teacher encoder and the student encoder, and the output of the teacher encoder is used as the pseudo-label available to the student encoder, and the parameter update of the teacher encoder is guided by the exponential moving average of the output of the student encoder; a small data set with GGO classification labels is used to fine-tune and train the fine-tuning encoder. It can effectively improve the classification accuracy and classification efficiency of the IA type of GGO imaging features.

The present invention discloses a method for synchronously extracting and detecting 59 endocrine disruptors in amniotic fluid, belonging to the technical field of substance detection. The method includes the following steps: S1, adding β-glucuronidase to the amniotic fluid sample and incubating; S2, adding a stable isotope-labeled internal standard to the amniotic fluid sample to be detected in step S1; S3, adding the amniotic fluid sample to be detected in step S2 to an activated HLB extraction column to remove impurities and extract the target analyte; S4, eluting the extraction column to obtain an eluate, blowing it to near dryness, and adding a reconstitution solvent to obtain a sample to be detected; S5, using a liquid chromatography-mass spectrometry instrument to detect the sample to be detected obtained in step S4 to complete the detection and analysis of endocrine disruptors. The detection process is carried out simultaneously in positive ion mode and negative ion mode. This method realizes the synchronous extraction, detection and analysis of at least 59 typical environmental endocrine disruptors in amniotic fluid, which is simple and convenient, and solves the problems of large sample consumption and long time consumption.

The present invention discloses a method, system, storage medium, and terminal for unbalanced data regression, belonging to the field of data regression technology. The method includes: constructing a multi-granularity space generation model based on grid division, dividing similar samples into the same granularity space; independently oversampling in each granularity space, wherein the oversampling range is defined by combining spatial importance and dynamic neighborhood radius. In scenarios with large sample sizes, the present invention comprehensively considers data flow and imbalance issues, improving processing efficiency while taking into account the overall distribution of data, reducing the interference of noise samples, and retaining the continuity and intrinsic correlation of samples, providing new ideas for the field of unbalanced data stream regression.