126 results about "Distance matrix" patented technology

The invention discloses a data dynamic partition storage method and system based on adaptive clustering, and relates to the field of data processing. The method comprises the following steps: S1, extracting multi-scale geometric features of a high-dimensional data set, calculating a local curvature and generating a curvature feature matrix; s2, constructing a feature distance matrix and a similar matrix based on the curvature feature matrix, and generating a low-dimensional embedding matrix; s3, executing self-optimization clustering according to the low-dimensional embedded matrix, determining a cluster number through singular value distribution, and generating an initial cluster; s4, calculating a stability factor of each cluster, and triggering cluster splitting or merging operation according to a threshold value to form updated cluster division; and S5, performing incremental processing on newly added data points, obtaining a new point local curvature through local curvature gradient correction, mapping the new point local curvature to a low-dimensional space, dynamically deciding affiliation based on a cluster radius, and updating partitions. Through multi-scale feature extraction, self-optimization clustering and incremental updating mechanisms, high-dimensional data clustering precision, robustness and calculation efficiency are improved.

The invention belongs to the technical field related to data processing, and provides a brain network analysis method and system based on multi-network collaborative topology analysis in order to solve the problem that an analysis result is unstable due to threshold selection subjectivity in existing Alzheimer disease brain network analysis. Extracting brain regions corresponding to the default mode network, the significance network and the execution control network, calculating correlation coefficients among different brain regions to construct a function connection matrix, and converting the function connection matrix into a distance matrix meeting complex construction requirements; constructing a Vietories-Rips complex based on the distance matrix, calculating a coherence group under each scale of the filtering sequence, and extracting topological features of a 0-dimensional Betti number and a 1-dimensional Betti number; according to the method, the Betti number is extracted, a curve that the Betti number changes along with the threshold value is drawn, the difference of the two groups in topological characteristics is compared, the specific topological biomarker related to the Alzheimer's disease is identified, and richer biomarker information is provided for early diagnosis of the Alzheimer's disease.

The invention discloses a translation memory bank retrieval and matching method based on semantic similarity, which relates to the technical field of computer-aided translation, and comprises the following steps: generating an internal edge and a cross-graph edge, calculating a one-time jump success probability, calculating edge and cross-graph edge cost based on cleanliness, correcting by using the one-time jump success probability to obtain an edge weight; generating a weighted graph, constructing an initial priority queue, searching by using a shortest path to obtain a distance matrix, calculating density based on a fuzzy similarity matrix, screening the density to obtain a genealogy center, calculating proximity, and generating candidate translations in combination with the distance matrix; static and dynamic fragment sets are constructed through a dependency syntax and BERT-NER, semantic matching precision and noise robustness are improved, a weighted graph is constructed in combination with cleanliness and transfer factors, shortest path matching is carried out, and the retrieval recall rate and quality of technical translation are improved.

The invention relates to a gearbox fault diagnosis model construction method based on metric guide graph comparative learning, and belongs to the field of gearbox fault diagnosis model construction. The method comprises the following four core stages: firstly, carrying out frequency domain conversion and normalization on vibration signals of the gearbox to generate a node characteristic matrix; secondly, a cosine distance and an Euclidean distance are fused to construct a mixed distance matrix, and a fault diagnosis graph is generated based on a K-nearest neighbor algorithm; then unsupervised graph comparison pre-training is realized through graph data enhancement and a dynamic graph attention network (DGAT); and finally, weak supervision fine tuning is carried out by using a small number of marked samples to complete construction of the gearbox fault diagnosis model. According to the method, the construction of the high-precision gearbox fault diagnosis model can be realized in a scene with extremely few marked samples (1-10 samples per class), and the method is suitable for planetary gearbox health monitoring in the fields of wind turbines, helicopters, hybrid electric vehicles and the like.

The invention discloses a micro-grid cluster dimension reduction method and system based on scene self-adaption and topology maintenance. Aiming at different types of energy, constructing a steady-state characteristic index system and decoupling the steady-state characteristic index system into a plurality of groups of characteristics; the method comprises the following steps: clustering system time series data, constructing an operation state manifold diffusion matrix, clustering by using a fuzzy clustering method based on diffusion distance, and outputting an optimal scene division result; constructing a Wasserstein distance matrix by using the steady-state characteristic index of each type of energy, defining a scene objective function and solving an optimal index weight; and calculating a scene membership degree and a fusion index vector weight according to the real-time operation state, weighting the steady-state characteristic index system, and outputting a weighted characteristic matrix. And performing multi-scale topology analysis on the weighted feature matrix, constructing a topology stability constraint, and aggregating the power grid equipment by adopting a hierarchical clustering method based on the topology stability constraint. According to the technical scheme, the internal evolution rule of the operation state can be accurately captured, so that the aggregation result better meets the actual operation requirement.

The invention provides a power distribution network topology construction splicing method based on recursive grouping and a graph attention network. The method comprises the following steps: acquiring electrical attribute information of node equipment in a power grid by using a monitoring sensor; recursive grouping is carried out, nodes and branches in the feeder line of the power distribution network are identified by using the constructed impedance matrix and distance matrix, an initial topological graph is constructed, and a node feature matrix and an initial adjacent matrix are obtained; a GATv2 graph attention network is introduced, dynamic attention calculation is performed by using the node feature matrix and the initial adjacent matrix, and a final topological graph of the feeder is generated; a landmark matching method and Pearson's correlation coefficient calculation are adopted, and boundary points of each independent topology are identified and matched; performing accurate alignment on the topological graph generated by each feeder line by using a graph embedding synchronization technology, and finally performing global topology processing; and carrying out topology verification on the constructed global topological graph. Construction and efficient splicing of line topologies in different areas of the power distribution network are achieved, and efficient operation and intelligent development of a modern power distribution network are supported.

The invention provides a space dimension reduction method for dynamic trajectory analysis, and belongs to the field of data analysis and machine learning. The method takes dynamic trajectory data as Riemannian function type data to carry out nonlinear full dimension reduction, and comprises the following steps of: 1, converting the Riemannian function type dynamic trajectory data into Euclidean space function type dynamic trajectory data; 2, constructing a Gram matrix of time data and a distance matrix of measurement space data for the Euclidean space function type dynamic trajectory data; and step 3, constructing a dimension reduction function to carry out nonlinear dimension reduction on the Euclidean space function type dynamic trajectory data. According to the method, dynamic trajectory data with a complex geometric structure can be effectively processed.

The invention discloses a beam moving gantry crane winch fault diagnosis method, and relates to the technical field of fault diagnosis, and the method comprises the steps: carrying out the wavelet decomposition and reconstruction of a gantry crane winch bearing monitoring signal, and generating denoising data; respectively inputting the de-noised data into a first feature extraction path and a second feature extraction path, generating a symbolized embedded vector by the first feature extraction path through standardization, symbolization and phase-space reconstruction processing, and calculating a relative entropy feature value of mode distribution; the second feature extraction path calculates a yoke divergence matrix and generates a dependency feature value by constructing an original embedding vector and a delay embedding vector; constructing two-dimensional feature vectors, and calculating the Euclidean distance of the feature vectors between the samples to generate a distance matrix; performing multi-dimensional scale transformation on the distance matrix to generate low-dimensional space distribution coordinates of the sample; and determining a fault category based on distance matching between the low-dimensional space distribution coordinates and a known fault category space region. And accurate diagnosis of different faults is realized.

The present application belongs to the technical field of data processing, and particularly relates to a method and system for dynamic monitoring of energy carbon based on multi-source data fusion, which comprises the following steps: collecting high-frequency energy data and low-frequency emission data of a target area, performing frequency-raising processing on the low-frequency emission data by using an interpolation algorithm, and performing standardization processing on the two types of data to obtain energy sequences and emission sequences with unified time resolution; calculating the causal fluctuation confidence of the energy sequences at each time, and constructing a cumulative distance matrix in combination with the causal fluctuation confidence; searching for an optimal alignment path of the energy sequences and the emission sequences by using a dynamic time warping algorithm, and calculating a local slope index at each path point; dynamically adjusting the observation noise covariance of Kalman filtering according to the local slope index, performing smoothing correction on the aligned emission data, and outputting the energy carbon monitoring result. The present application solves the problem of distortion in calculation of carbon emission intensity caused by phase misplacement of multi-source data.

The present application relates to the technical field of position estimation, in particular to a sea mobile target motion prediction method and device based on limited prior information. The method comprises the following steps: acquiring prior information; constructing a tangent plane coordinate system based on the prior information; calculating a grid side length; determining a unit time step; constructing a voyage transfer probability density distribution function and a heading transfer probability density distribution function; determining a task area, performing grid processing on the task area based on the grid side length, performing discrete point processing on each grid, and acquiring a discrete point one-dimensional vector of each grid; constructing a transfer distance matrix based on a starting grid and a target grid, constructing a transfer angle matrix based on the starting grid, the target grid and a destination point, and acquiring a transfer probability matrix based on the transfer distance matrix and the transfer angle matrix. In this way, the target key area can be quickly searched through observation under limited prior information, thereby improving the target search perception efficiency.

The application discloses a biped robot form optimization method based on metric index identification variable additive grouping, which comprises an additive structure grouping stage and a Bayesian optimization stage based on additive decomposition; in the additive structure grouping stage, first, the interaction strength between variables is calculated through the designed variable interaction metric index, and the interaction strength is converted into a distance matrix; subsequently, using the distance matrix, a K-Means clustering algorithm with capacity limitation is used to cluster high-dimensional variables to obtain grouping, so that an additive grouping structure is obtained; in the Bayesian optimization stage based on additive decomposition, an additive kernel GP proxy model is constructed based on the obtained additive grouping structure, the next evaluation point is obtained by maximizing the obtained function, and the form parameters of the evaluation point are used as the initial form parameters of the biped robot for final application. Compared with the existing method, the application is more simple and robust, and by limiting the maximum dimension of each sub-group, the optimization efficiency and grouping identification accuracy are balanced.

The invention provides a UUV cluster positioning method based on merging enhanced multi-dimensional scaling analysis, and relates to the technical field of cluster positioning, and the method comprises the steps: obtaining coordinate information of anchor nodes and an observation distance matrix of a node group, and constructing different reference groups based on reference points; for the same reference group, combining the reference group with nodes which do not belong to the reference group in the node group to obtain sub-combinations, calculating relative coordinates of the nodes in each sub-combination by using an MDS algorithm, and then performing rigid body registration to obtain global relative coordinates corresponding to the reference group; performing alignment processing on the global relative coordinates based on the coordinate information of the anchor nodes to obtain an absolute coordinate set corresponding to the reference group; and performing optimization based on each absolute coordinate set to obtain a positioning result of the to-be-positioned node. According to the method, the position information of the reference point is fully utilized, the ranging error and the channel noise are remarkably suppressed, the positioning effect is improved, and the problem of performance degradation of the traditional MDS in a low-connectivity scene is solved.

The invention discloses a matching-based privacy-preserving average causal effect estimation method and system. The method comprises the following steps: firstly, training a logistic regression model, and injecting noise into a model weight; thirdly, calculating the tendency score of each sample by using the disturbed model, and adding noise; then calculating the distance between different samples according to the tendency score of each sample after disturbance, and sorting the distance; thirdly, matching is achieved based on the ranked distance matrixes, so that the potential intervention result of each sample under disturbance setting is calculated, noise is further injected into the sum of the potential intervention results of all the samples, and estimation of the average causal effect is achieved based on the sum of the potential results of disturbance.

This invention proposes a basic matrix fitting method based on residual sorting and sparsity constraints. The fitting method generates an effective distance matrix for a dataset containing multiple model instances and a high proportion of outliers, specifically including the following steps: Step S1: Set the input data and the number of model instances. S Number of samplings M Step S2: By improving the algorithm AGS, sampling weights and information theory principles are used to select effective data, and then a subset of data is selected from the effective data to generate... M One model hypothesis; Step S3: Use the generated model hypothesis. M The model assumes calculation of the discrimination distance matrix; Step S4: Calculate the sparse distance matrix using information theory; Step S5: Apply parameters... S The class labels are obtained by performing spectral clustering; this invention can effectively combine the advantages of residual sorting and sparsity constraints, and can efficiently generate more accurate distance matrices for data containing multiple model instances and a high proportion of outliers.

The application relates to the technical field of dam body leakage detection, in particular to a reservoir dam body early leakage detection method, equipment and system, the method comprising the following steps: obtaining synchronous aligned reservoir water level and pore water pressure, extracting pressure response time lag and pressure response amplitude ratio under a dominant frequency by using frequency domain analysis, constructing a hydraulic retardation modulus representing medium properties, calculating the shortest seepage path length from each measuring point to the reservoir water boundary according to a dam geometric structure model and constructing a spatial adjacency distance matrix, performing linear trend fitting on the pressure response time lag and the seepage path length to generate data reliability weights, calculating the distribution recovery cost required for transforming the measured retardation distribution vector into a reference path distribution vector, and performing trend monitoring on the distribution recovery cost at continuous multiple time points to determine abnormalities. The application aims to extract the medium property change characteristics caused by early small reservoir dam body leakage, and realize spatial positioning of the leakage source position.

The present application relates to the field of automation, and discloses an anti-pinch early warning method and apparatus, and an electronic device. The method comprises: determining distance values between sensing points of a sensor module and the sensor module, and constructing a target distance matrix on the basis of the distance values; acquiring an operating state value of a movable end, and determining a target template matrix on the basis of the operating state value; performing differential calculation on the target distance matrix and the target template matrix according to a preset rule to obtain a target difference matrix, and grouping matrix element values in the target difference matrix on the basis of connectivity of the matrix element values in the target difference matrix, so as to obtain Q matrix element value sets; and determining the number of elements in each matrix element value set, and determining an anti-pinch early warning result on the basis of the number of elements. The present application solves the problem in the related art of a high risk of a user being pinched due to the fact that an anti-pinch operation can be started only upon detecting that a movable end of a target device is in contact with a foreign matter.

The application discloses a multi-target path optimization method and device and a medium, and relates to the technical field of path planning. The method comprises the following steps: constructing a weighted graph model of a region to be planned and verifying an adjacency matrix to obtain verified weighted graph data; calculating the shortest path distance between all node pairs to form a full-source shortest path distance matrix and a predecessor record set, performing connectivity detection and feasible region pruning processing on unreachable node pairs; generating an optimized access sequence covering all target nodes based on the processed full-source shortest path metric data and a nearest neighbor greedy strategy, accumulating a composite path cost, performing a feasibility test on a time window, a capacity and a risk constraint to generate an initial path sequence and accumulated cost data; expanding adjacent node pairs in the initial path sequence into specific executable paths on the original graph according to the predecessor record set, and performing neighborhood optimization on the executable paths by using a local search operator, and outputting an optimized final path scheme and an accumulated cost report.

The invention discloses a weak sound source sound pressure level calibration method based on an improved DAS beam, relates to the technical field of sound source positioning, and is used for solving the problem of repeated weak sound source positioning. According to the method, a distance matrix is constructed based on array geometric parameters, and a cross spectrum matrix and a steering vector are respectively calculated for a plurality of target frequencies in combination with a near-field propagation model to obtain a multi-frequency acoustic image; and performing normalization and relative threshold side lobe suppression on each frequency sound image result, superposing and highlighting a real sound source area on a scanning grid, suppressing a side lobe irrelevant to a position, and utilizing a sound pressure level calibration model obtained by training reference data to convert a sound image output quantity into a real sound pressure level, so as to realize spatial positioning and quantitative sound pressure evaluation of multiple sound sources. The weak sound source can still be clearly distinguished under the background that the sound pressure level difference of the strong and weak sound sources is large and the noise is complex, and the sound source positioning precision and the engineering detection reliability are improved.

The invention relates to a moving load applying method for a non-uniform discretization finite element model, which comprises the following steps: constructing the non-uniform discretization finite element model, determining the node number of each finite element node in the non-uniform discretization finite element model and the corresponding space coordinate of each finite element node, and enabling the unit length acted by the moving load to be in non-uniform distribution; calculating the spatial position of the jth moving load under the ith time step; judging the unit where the jth moving load is located under the ith time step, obtaining the node number of the left node of the unit where the moving load is located under the time step, defining the node number as a track matrix, calculating the distance between the moving load and the left node under the time step, and defining the distance matrix; repeating the steps to obtain a track matrix and a distance matrix of all moving loads in the total time step; and calculating the equivalent node force borne by the left node and the right node of the unit when the moving load acts on the unit. In a full-time-history analysis range, the moving load is dynamically tracked and accurately positioned so as to be accurately applied.

This application relates to a marker point matching method, splicing method, apparatus, device, and storage medium. The method includes: obtaining a current frame marker point set and a buffer frame marker point set; the current frame marker point set includes current frame marker points; the buffer frame marker points in the buffer frame marker point set are selected from global marker points; obtaining a first distance matrix corresponding to the current frame marker points and a second distance matrix corresponding to the buffer frame marker points; based on the first and second distance matrices, obtaining a pair of matching points with the same name for the current frame marker points and the buffer frame marker points; the matching point pair includes the same-name points of the current frame marker points; updating the buffer frame marker points based on the same-name points of the current frame marker points to obtain updated buffer frame marker points; the updated buffer frame marker points are used for matching with marker points in the next frame. This method can shorten the marker point matching time, thereby improving real-time performance.

The invention discloses an intelligent storage multi-AGV cooperative path planning method, device and system and a medium, and relates to the technical field of path planning. The path planning method comprises the following steps: acquiring carrying task information of a target AGV so as to determine an operation point set which needs to be accessed by the AGV; warehouse channel topological information is obtained, and the shortest driving distance between any two operation points is calculated to obtain a distance matrix; and constructing a traveling salesman problem instance by taking the operation point as a node and the distance as an edge weight. Obtaining m nearest neighbor nodes of each node in the distance matrix in the traveling salesman problem instance and a local neighborhood matrix of each node; and inputting the local neighborhood matrix into a pre-trained anchor point prediction network to predict whether each node is an anchor point or not, thereby dividing all nodes into an anchor point set and a non-anchor point set. And based on the distance matrix, the anchor point set and the non-anchor point set, constructing a planning path through an anchor point insertion algorithm. And converting the planned path into a transport path scheduling instruction of the AGV.

The application relates to a kind of distribution network multi-branch line fault accurate positioning method based on travelling wave positioning, comprising: 1, build double-end travelling wave synchronous acquisition device, collect line travelling wave signal and carry out denoising and time synchronization processing;2, carry out distribution network dynamic topology-travelling wave joint modeling, calculate and establish fault distance matrix using Dijkstra algorithm;3, the fault distance matrix is screened by density clustering algorithm, the data affected by time error is screened out, the data after screening is substituted into optimization model, and the basic data after error correction is calculated;The effective coordinate cluster obtained by clustering is calculated, the branch node coordinates are proposed, and the centroid and node euclidean distance are calculated, finally, the euclidean distance is compared with threshold value, less than, is branch fault, otherwise, is trunk fault, the double-end ranging formula and the trunk-to-branch node formula are used to calculate the results of trunk and branch fault respectively, and the correction of fault positioning result is realized.The application realizes the accuracy of fault positioning.

The application relates to the technical field of data processing, and discloses a city-level intelligent community multi-source heterogeneous data standardization management method, which comprises the following steps: acquiring heterogeneous data messages and generating a structure topology fingerprint, performing differential processing according to a fingerprint comparison result; when the comparison result is inconsistent, extracting time sequence statistical features of unknown fields to construct a physical feature vector, and calculating the topological structure correlation among the unknown fields to construct a topological distance matrix; based on the physical feature vector and a standard field reference vector, combining the topological distance matrix to construct and solve a global mapping loss function, and establishing the mapping relationship between the unknown fields and the standard fields; generating a streaming mask according to the mapping relationship and a message network communication protocol type, and extracting and storing standardized data based on the mask. The application can perceive message structure changes through a structure fingerprint, and can constrain a mapping relationship by combining space-time joint features, so that the accuracy of heterogeneous data mapping conversion and the system processing efficiency are improved.

The invention discloses a packaging process parameter data processing and consistency evaluation method and system, and relates to the technical field of data processing and quality consistency evaluation. According to the method, multi-source data fusion is carried out on the side wall morphology features after wafer scribing and the infrared thermal image dynamic features in the parallel seam welding process, and refined space-time two-dimensional fragmentation processing is carried out on the seam welding process, so that the problem that the packaging process state cannot be comprehensively reflected by a traditional single index is effectively solved; according to the method, multi-dimensional characteristic parameters are mapped to a unified coding interval and principal component dimensionality reduction is carried out to generate a fingerprint code capable of representing the comprehensive process state of a device, and then statistical analysis and heat map visualization of an Euclidean distance matrix are utilized to realize objective quantification of process consistency in batches and visual positioning of systematic deviation.

A method for determining a position of an object includes determining a first estimated location of the object based on a distance matrix, a range matrix, and a weight matrix, updating the weight matrix based on distance related factors and ranging technique related factors, and determining a second estimated location based on the distance matrix, the range matrix, and the updated weight matrix. The distance matrix includes distances between the object and one or more anchors and the range matrix includes ranging measurements for ranging communications between the object and the one or more anchors.

The application discloses a similar text retrieval method based on vector representation. The method first processes the text at the article level by using the word embedding + pooling method, obtains the text vector representation through Bert coding, constructs all text vector representations into a ball tree index by using the ball tree method, performs product quantization on all text vector representations and ball tree center points, and builds a PQ index; during retrieval, the query text is also converted into a vector representation, a distance matrix is calculated through product quantization, KNN retrieval is performed based on the distance matrix and the ball tree index structure, and k similar texts are obtained. The similar text retrieval method based on vector representation constructed by the application supports long text retrieval, can accurately and quickly retrieve massive text vector representations, has good effect on complex or multi-semantic text queries, is very friendly and helpful to users who are not familiar with the jargon of a certain field or want to do exploratory search, and finally, the search is more comprehensive and accurate compared with the IndexIVFPQ method in Faiss.

The application provides a kmeans log classification method based on adaptive radius center selection, and relates to the technical field of big data. The method comprises the following steps: fusing log samples until the number of log samples in a target sample set meets a first preset number condition, and constructing a first distance matrix according to the cosine distance of each log sample; storing a first center point obtained according to the first distance matrix into a center set, if a preset condition is triggered, continuing to fuse and update the target sample set until the number of log samples meets a second preset number condition, constructing a second distance matrix according to the fused and updated target sample set, continuing to obtain a second center point from the fused and updated target sample set according to the second distance matrix, and storing the second center point into the center set until the center point in the center set is K for Kmeans clustering. The scheme improves the quality of initial centroid selection, thereby improving the Kmeans clustering effect.

The invention discloses a rapid equivalent analysis method and system for a heat dissipation cold plate based on a Kriging agent model, and belongs to the field of heat dissipation design of electronic equipment, and the method comprises the following steps: obtaining a geometric model of the heat dissipation cold plate, and carrying out grid division and simulation to obtain design points of a grid model of the heat dissipation cold plate; after the key simulation data and the corresponding temperature response values serve as a training set to be normalized, a distance matrix is obtained through calculation, a regression matrix and a correlation matrix are constructed in combination with the key simulation data of the design points, finally, parameters of a Kriging agent model are optimized in combination with boundaries of the key simulation data, and the Kriging agent model is obtained. And obtaining a Kriging agent model for rapid equivalent analysis of the heat dissipation cold plate, and inputting a to-be-tested design point into the Kriging agent model to obtain a predicted temperature response value. According to the method, the processes of geometric modeling, grid division and numerical solution do not need to be carried out again after subsequent parameter adjustment, and the efficiency of equivalent analysis of the heat exchange performance of the heat dissipation cold plate is improved.

This invention discloses an unsupervised image defect detection method, apparatus, and medium based on knowledge distillation. The method includes: constructing a teacher model, training it using a dataset unrelated to the domain of the defect to be detected, and then providing knowledge guidance beneficial to defect detection to a student model; constructing a student model to extract multi-scale feature maps of the image; using the student model to construct a multi-scale feature set of defect-free images, obtaining a student feature map set; establishing a pixel-level defect detection model set using the student feature map set; receiving an image, a pixel-level defect threshold, and an image-level defect threshold sent by a user terminal; calculating the distance between the multi-scale feature maps and the pixel-level defect detection model set, obtaining a distance matrix; and determining whether the image sent by the user terminal has a defect by using the relationship between the distance matrix and the pixel-level and image-level defect thresholds. This invention only requires defect-free samples to solve the problems of insufficient defect samples and defect-free samples.

The invention belongs to the technical field of water supply network monitoring, and particularly relates to a water supply network leakage detection method based on a Kriging interpolation method and a CatBoost model, and the method comprises the steps: firstly constructing a network hydraulic model, and calculating an inter-node Euclidean distance matrix; a hydraulic model is used for simulating multiple times of leakage, and pressure data of pressure value monitoring points and corresponding leakage labels are collected to form training samples; based on a Kriging interpolation method, fusing the pressure data and the distance matrix, estimating a pressure value of a non-monitoring point, and constructing a pressure feature set covering nodes of the whole network; performing importance sorting and screening on the features to obtain an optimal feature subset; combining the data with leakage labels to form a training sample matrix, and inputting the training sample matrix into a CatBoost model for training; and utilizing the trained model to realize the identification of the leakage area and grade according to the real-time pressure data. According to the invention, based on the pressure data of a small number of monitoring points, the independent metering partition where leakage occurs can be positioned more accurately, and the severity of the leakage can be judged.