62 results about "Hierarchical cluster algorithm" patented technology

The invention discloses a multi-disturbance scene-oriented wind turbine generator coupling modeling and clustering analysis method, and relates to the technical field of wind turbine generator running state monitoring. According to the method, running data of a unit in mechanical, electrical and wind regime physical domains are synchronously acquired through multi-source data, and a cross-domain coupling correlation tensor is constructed based on a phase-space reconstruction theory; extracting a coupling feature representing a multi-physical domain dynamic interaction relationship; forming a fusion scene feature vector by combining the shape feature and the topology invariant of the disturbance signal; based on a self-adaptive hierarchical clustering algorithm of a shape sensing distance, accurate classification of multiple disturbance scenes and typical scene center identification are realized; carrying out clustering result verification and scene identification; a self-adaptive response strategy is generated based on typical scene features, and intelligent optimization and safety early warning of the unit operation state are achieved; and the state sensing and self-adaptive regulation and control capabilities of the wind turbine generator in a multi-disturbance scene are effectively improved.

The invention discloses an industrial and commercial time-of-use electricity price dynamic adjustment model considering an electric power spot market. The method comprises the steps of obtaining a historical load curve of a preset category; dividing the historical load curve into a first time period group based on a first preset quarter and a second time period group based on a second preset quarter based on a hierarchical clustering algorithm and by combining a load weighted Euclidean distance and a Ward connection method according to the historical load curve; obtaining a clustered load curve according to the first time period group and the second time period group; determining load characteristics according to the clustered load curve; adjusting a load elasticity coefficient according to the load characteristics and historical load data; constructing an optimized load model according to the load elastic coefficient; inputting the optimized load model into a single-target artificial hummingbird algorithm, and determining the electricity price of each time period; and determining a final electricity price scheme according to the electricity price of each time period. The invention solves the problem of insufficient seasonal load adaptability of the current power system.

The invention discloses a data aggregation processing method and device, computer equipment and a storage medium, belongs to the technical field of big data, and is applied to data aggregation in the financial field. The method comprises the following steps: firstly, grouping dimension fields by utilizing a mapping rule and a hierarchical clustering algorithm; and secondly, dynamic grouping mapping is established, one-time scanning is performed on the source data table, and each data record is directly attributed to a corresponding grouping structure, so that the data processing efficiency is improved. Meanwhile, a correlation index of the index and the data partition is constructed based on a Hash mapping algorithm. Under the support of an indexed data structure, the aggregation calculation function can efficiently execute operations such as summation and averaging, and rapid aggregation of multi-dimensional data is realized. Generally speaking, according to the method, the correctness of an aggregation result is ensured, meanwhile, the calculation and storage pressure of the system is remarkably reduced, the timeliness and expandability of data analysis are improved, and the method is particularly suitable for real-time or quasi-real-time multi-dimensional analysis requirements in a large-scale business scene.

The invention discloses an event ontology mode construction method and device under the assistance of a large model, and belongs to the technical field of ontology construction and evolution.The event ontology mode construction method comprises the steps that a large language model is used for precisely screening corpus metadata and automatically extracting event elements; the method comprises the following steps: converting an event quintuple into semantic, time sequence and relation vectors through a deep semantic representation model, dynamically fusing time sequence and relation characteristics by taking the semantic vector as a dominant, and iteratively merging the dynamic event vectors by adopting a time sequence-relation enhanced hierarchical clustering algorithm to construct a clustering tree with a father-child hierarchical relation; generating general Chinese class names for non-leaf nodes through a large language model, and forming an event ontology skeleton which is easy to understand; and finally, the ontology skeleton is mapped into OWL ontology classes, individuals and object attributes conforming to the RDF / OWL specification, a standardized ontology file is generated, automatic construction and output of the event ontology mode are achieved, and the automation degree and normalization of event ontology construction are improved.

The invention discloses a sales demand analysis method and system based on AI and portraits. The method comprises the steps of collecting multi-source heterogeneous data such as basic information, transaction records and online behaviors of customers; according to the method, a self-developed large model of the Chinese human life research and development center is adopted, implicit requirements in a customer behavior sequence are deeply mined through an attention mechanism, a hierarchical clustering algorithm is used for generating a tree-shaped clustering graph based on a Ward connection method, and high-value features with variances larger than 0.8 and absolute values of correlation coefficients larger than 0.5 are screened out in combination with a rule engine; it is ensured that the extracted features have statistical significance and are directly associated with business targets, invalid feature redundancy is avoided, the efficiency and accuracy of subsequent portrait construction are improved, product recommendation priorities are determined through a decision tree algorithm, and a Q-learning framework is combined to optimize communication verbal skills and cross sales strategies, so that the sales strategies meet the current demands of customers, and the customer experience is improved. Potential requirements are guided, and the conversion rate and the customer satisfaction degree are both improved.

The present application relates to the technical field of mine risk assessment, and discloses a mine operation risk grade assessment system. The geological data acquisition module of the system identifies potential rock layer instability risk areas through multidimensional sensing monitoring and hierarchical clustering algorithm; the mine image analysis module labels thermal anomaly feature areas with the help of thermal infrared image acquisition and feature separation and pattern recognition; the risk area synthesis module integrates the two areas to generate a comprehensive risk monitoring area; the structure displacement evaluation module analyzes rock layer displacement vector changes to determine abnormal fluctuation degree; the ground pressure interaction evaluation module analyzes rock layer stress and groundwater pressure interaction to evaluate geological stress field coupling imbalance index; and the risk grade determination module determines the risk grade classification of the comprehensive risk monitoring area according to the above evaluation results, which can comprehensively and accurately evaluate mine operation risk and ensure operation safety. The system improves the accuracy of rock mass stability judgment and can detect potential rock layer instability hazards earlier.

The application relates to a kind of millimeter wave radar multi-vehicle tracking methods based on traffic scene, comprising: the original echo data of millimeter wave radar is preprocessed, and a plurality of frames of point cloud data are obtained;Each frame of point cloud data is clustered using a hierarchical clustering algorithm based on target information;Rectangular association wave door is set in combination with standard lane width and safe following distance, and the rectangular association wave door is corrected according to the historical information obtained by target tracking;All target centroid points in the corrected rectangular association wave door are fused according to probability;The existing track associated with the equivalent centroid point and the existing track not associated with the equivalent centroid point are updated, and a plurality of updated tracks are obtained;For the target centroid point not assigned to the existing track, the logical method is used for track initiation, and a plurality of stable starting tracks are obtained;The target tracking track is obtained by managing the track. The method combines the characteristics of radar measurement distribution and motion mode of vehicle target, and realizes the robust tracking of multi-vehicle target.

The present invention relates to a measurement method for high-density three-dimensional composite urban form characteristics, and a computing apparatus and a storage medium. The measurement method comprises: delineating a three-dimensional identification boundary, and acquiring raw map data; performing screening processing on the raw map data, and performing three-dimensional modeling within the identification boundary; calculating three-dimensional characteristic measurement indices; selecting three-dimensional base coverage rate and three-dimensional base plot ratio to construct a coordinate system, and classifying sample indices into hierarchies according to distribution characteristics; using a hierarchical clustering algorithm to constantly perform iteration until all data is combined into one category; and storing the data, building a three-dimensional urban database, and keeping the data updated. The present invention is aimed at the form characteristics of three-dimensional composite urban development during the renewal of high-density cities, and solves the problems in the prior art of it not being possible to precisely express three-dimensional urban forms and it also not being possible to perform effective quantitative analysis on the three-dimensional urban forms during urban design.

The application discloses a patent theme graph generation method and system based on deep semantic hierarchical clustering, comprising: obtaining a patent literature set PT to be generated into a patent theme graph; using a patent deep semantic representation model to perform semantic coding on each patent literature in the patent literature set PT, and obtaining a semantic representation vector matrix V={v1,...,v N}; inputting the semantic representation vector matrix V into a hierarchical clustering algorithm to obtain a hierarchical clustering tree structure corresponding to the patent literature set PT; generating a corresponding theme description for each non-leaf node on the hierarchical clustering tree structure; and combining the generated theme description and the hierarchical clustering tree structure into a patent theme graph with an upper and lower hierarchical structure. The patent theme graph generation method and system can help users quickly mine upper and lower hierarchical relationships of patent literatures at a theme level, meet the analysis needs of users for large-scale patent information, and improve the efficiency of patent analysis of users.

The application provides a prostate cancer biochemical recurrence prognosis risk prediction model based on fatty acid metabolism and cancer cell stemness genes and a construction method thereof, wherein the construction method comprises the following steps: S1: data collection: obtaining prostate cancer sample transcriptome data with biochemical recurrence information from a database, and dividing the data into a model training set and a model test set; S2: stemness score analysis; S3: fatty acid metabolism score analysis; S4: based on the analysis results of S2 and S3, identifying a co-expression gene module related to fatty acid metabolism and stemness characteristics in prostate cancer by a co-expression similarity algorithm and a hierarchical clustering algorithm, and obtaining a fatty acid metabolism and stemness-related gene set; S5: constructing a prostate cancer BCR prognosis risk prediction model; S6: constructing a nomogram model; S7: extracting RNA of a to-be-tested sample, constructing a cDNA library, quantifying the expression of the above genes, and calculating the prognosis risk level of prostate cancer through the expression level.

The invention discloses a KRAS protein conformation change simulation method and system based on multi-sample analysis, and belongs to the technical field of multi-sample analysis. The method comprises the following steps: collecting multi-sample molecular dynamics simulation data, calculating energy difference between conformations, and constructing an energy difference mapping matrix; then, based on a joint criterion of an energy difference mean value and an energy gradient change rate, a hierarchical clustering algorithm is adopted to divide a conformation energy space into a main steady-state region, a metastable-state region and a transition-state region, and a conformation hierarchical tree structure is formed; and extracting a key channel and a stable sub-state of conformation transformation according to the relationship between the energy fluctuation characteristics and transformation paths of various intervals, thereby realizing quantitative description and energy path identification of the KRAS protein conformation change. According to the method, dynamic conformation relations under different energy levels are effectively distinguished, a new calculation and analysis means is provided for conformation regulation and control and drug binding site recognition of a complex biomolecular system, and the method has the advantages of being high in accuracy, high in universality and good in expandability.

The invention discloses a differentiated regulation and control method for irregularity of a large-span bridge track of a high-speed railway. The method comprises the following steps: integrating track irregularity historical detection data at different environment temperatures into multi-stage track irregularity historical detection data with environment temperature labels; establishing a multivariate empirical wavelet transform method, and obtaining a plurality of intrinsic mode components; defining the intrinsic mode component as an intrinsic mode, extracting multi-dimensional physical feature vectors of all the intrinsic modes to construct feature vectors, and performing unsupervised classification on all the intrinsic modes by using an agglomerated hierarchical clustering algorithm; component category division is carried out according to a classification result, and different differentiation regulation and control strategies are made according to different component categories; and when the orbit adjustment amount required by the classification exceeds the limit, establishing an optimization model containing the adjustment amount physical constraint and the smoothness expectation target, and solving the optimal orbit adjustment amount considering the current state and the future temperature adaptability.

This invention discloses an intelligent analysis method and system for underground structures based on multi-source geological data fusion, relating to the fields of engineering geology and geotechnical engineering. The method includes: S1, collecting raw geological data and applying a hierarchical clustering algorithm to group scale-scale fractures and interlayer slippage to obtain a preliminary stiffness distribution map of scale-scale rock mass units; S2, based on the preliminary stiffness distribution map of the scale-scale rock mass units, using finite element analysis to simulate the stress transfer path between units and determine the deformation mode and failure path of the scale-scale geological blocks. This intelligent analysis method and system for underground structures based on multi-source geological data fusion can accurately identify stress concentration, uneven settlement, and potential sliding surface locations in key areas such as weak interlayers in dam foundations, reducing judgment bias caused by empirical simplification. It provides more objective, stable, and repeatable analysis results for stability evaluation and risk prediction of underground engineering, and has good engineering applicability and promotional value.

The application provides an intelligent insurance evaluation monitoring and early warning system based on a whole process of logistics transportation, and relates to the technical field of logistics monitoring. The system comprises the following steps: collecting transportation data in the process of logistics transportation; preprocessing the transportation data; using a condensed hierarchical clustering algorithm to perform unsupervised clustering on the preprocessed logistics data, dynamically dividing the transportation risk types, and obtaining risk cluster labels; constructing an intelligent prediction model combining a lightGBM model and a neural network model to obtain an output risk probability distribution threshold value at a future t time; and judging whether real-time detection data exceeds the preset risk probability distribution threshold value, and if so, issuing an alarm and reminding. The intelligent prediction model combining the lightGBM model and the neural network model is used to predict the risk probability distribution threshold value at the future t time, so that the time and computing resources required for model training are reduced, and the efficiency of the whole logistics risk prediction work is improved.

The application discloses a fabric classification method and device based on color characteristics, and a storage medium. It relates to the field of data processing. The method comprises the following steps: printing a color standard on each fabric to be classified; using a spectrophotometric colorimeter to collect the CIE Lab color values of each color sample of the color standard on each fabric to be classified; constructing a curve corresponding to the ink of each color on each fabric to be classified; calculating the curve difference between the curves corresponding to the ink of each color between any two fabrics to be classified; obtaining the fabric difference value between any two fabrics to be classified; constructing a symmetric difference matrix; using a hierarchical clustering algorithm, based on the symmetric difference matrix, classifying the fabrics to be classified, and obtaining a classification result. Through the application, the problem of low efficiency of color management of fabrics due to the inability to classify fabrics based on the color characteristics of the reaction between fabrics and ink relying on artificial experience in related technologies is solved.

The invention relates to the technical field of 5G video customer service, and discloses an intelligent video customer service access model establishment method based on 5G. The method comprises the following steps: collecting real-time video stream data transmitted by a user terminal through a 5G network, extracting multi-dimensional session features such as facial expression change, voice intonation, gesture action and the like, and constructing a dynamic intention weight matrix in combination with a historical session record intention classification result; performing noise reduction on the video stream by using an adaptive filtering algorithm, extracting a key action frame and calculating a time sequence correlation degree; the dynamic intention weight matrix and the key action frame time sequence correlation degree are fused, and an enhanced user intention vector is generated; based on the matching degree of the vector and historical session record intentions, dividing user intention categories through a hierarchical clustering algorithm, and establishing a mapping relationship between the user intention categories and a customer service strategy; and finally, generating an access instruction according to the mapping relation, dynamically allocating optimal customer service resources, and adapting to 5G network characteristics.

The invention discloses an intelligent algae monitoring method and system based on a floating bed process. The method comprises the following steps: constructing a floating bed area map model; collecting multi-dimensional algae and environmental parameters before and after floating bed regulation; performing linear regression prediction on the serialized parameters to form new parameter data, and vectorizing the parameters to obtain algae and environment feature vectors; introducing a hierarchical clustering algorithm to evaluate regional feature similarity, and analyzing an algae reproduction and migration direction and an environment matching region; and comparing and analyzing the algae migration trend and the environment change before and after regulation, evaluating the floating bed regulation effectiveness and optimizing the floating bed arrangement scheme. According to the method, intelligent algae monitoring and water ecology multi-point trend prediction are achieved, the algae migration law is accurately recognized, the effectiveness of floating bed setting is scientifically evaluated, water body deterioration is effectively relieved, the water ecology regulation and control efficiency is improved, and early warning hysteresis is reduced.

The application provides a defense strategy determination method based on industrial control network flow data, comprising the following steps: firstly, obtaining at least one archived industrial control flow data of a target industrial control network node; obtaining flow service scenario vectors of a plurality of archived industrial control flow data based on a K-means clustering algorithm; for each of the archived industrial control flow data, determining a first risk coefficient of the archived industrial control flow data from a strategy table for storing a first risk defense strategy; then, performing a hierarchical clustering algorithm on the plurality of archived industrial control flow data to obtain a first risk classification vector; constructing a risk defense mapping library of the first risk defense strategy according to the first risk classification vector of each of the archived industrial control flow data and the service scenario to which each of the archived industrial control flow data belongs in the flow service scenario vector; and constructing a network attack defense mechanism of the target industrial control network node based on the risk defense mapping library. The above method can improve the security of the entire industrial control network.

The invention provides a method and a device for determining a geographic position of an optical network unit. The method comprises the following steps: determining the optical network unit; acquiring a plurality of pieces of undetermined geographical location data of the optical network unit; performing initial grouping on the optical network units according to the optical splitter, and clustering each group by using a hierarchical clustering algorithm to generate a corresponding final group; determining a plurality of undetermined geographical location data of the optical network units contained in the final group, detecting and removing abnormal geographical location data, and defining the remaining geographical location data as an effective geographical location data set; and carrying out average calculation on each effective geographic position data, and taking the result as the geographic position data of all optical network units in the final group corresponding to the effective geographic position data set. Therefore, the positioning accuracy of the optical network unit can be improved.

This invention discloses an artificial intelligence-based method for recommending buried pipeline layout, belonging to the field of intelligent recommendation technology. The method includes: converting the laying environment, construction purpose, and specific parameters of the buried pipeline into input vectors and inputting them into a pre-trained buried pipeline layout recommendation model to obtain an initial recommended layout scheme; mining historical new problems and solutions for each historical pipeline project during the pipeline laying process, forming several problem-strategy pairs; using a hierarchical clustering algorithm to perform cluster analysis on all problem-strategy pairs under all historical pipeline projects; calculating the similarity between the current buried pipeline laying environment and construction purpose and each historical pipeline project based on an autoencoder similarity algorithm to obtain the required retained problem-strategy pairs; and improving the initial recommended layout scheme according to the layout improvement type. The pipeline layout scheme is optimized and improved to generate a layout scheme that meets actual needs while also considering potential risk prevention.

The invention belongs to the technical field of traffic geographic information analysis, and discloses an urban crowd activity and space structure dynamic deduction method and system based on single traffic flow, a terminal and a storage medium. Identifying a spatial distribution mode through a flow similarity measurement method in combination with a hierarchical clustering algorithm; carrying out quantity aggregation on the identified spatial distribution mode according to grid units, and constructing a spatial grade distribution model; combining the urban built environment features with spatial grade distribution in the spatial grade distribution model, and constructing a comprehensive data set suitable for machine learning regression modeling; and based on the comprehensive data set, in combination with a machine learning model interpretation method of a game theory, quantitatively analyzing a nonlinear influence mechanism of urban built environment characteristics on different spatial distribution modes, and deducing internal relevance between urban crowd activities and spatial structures. The dynamic evolution process of the urban space structure is comprehensively realized.

The invention discloses a terminal area approach mode mining method based on ASM-HAC. The method comprises the following steps: acquiring ADS-B data of an airport terminal area; the track features are screened, and the correlation degree between the track features is analyzed in combination with a Pearson's correlation coefficient; obtaining a track distance matrix by using a multi-dimensional dynamic time warping method under weighted Euclidean distance; improving a Gaussian kernel function by using adaptive parameters to obtain an ASM track similar matrix; a Laplacian matrix and a feature gap method are introduced to improve an HAC hierarchical clustering algorithm, and after the improved HAC hierarchical clustering algorithm is applied to an ASM track similar matrix, an optimal track clustering result is obtained; and evaluating the rationality of the optimal clustering number by adopting a contour coefficient. According to the method, the problem that an original hierarchical clustering method excessively depends on manual intervention to obtain the optimal clustering number is solved, the similarity measurement requirement of a track dense region and a track sparse region is balanced, the clustering process better fits the space-time heterogeneity of track distribution, a controller is assisted in mastering the characteristics of different approach modes, and control decision making is facilitated to be completed.

An example method includes initiating training of an hierarchical clustering algorithm using training data. The method further includes determining a first factor, the first factor being a number of analyzed nodes compared to a number of discovered nodes. The method further includes determining a second factor, the second factor being a first time elapsed compared to a first typical training duration for environments with a data set size substantially similar to a data set size of the training data. The method further includes determining a third factor, the third factor being a second time elapsed compared to a second typical training duration for environments with a data having a uniformity substantially similar to a uniformity of the training data. The method further includes estimating a progress of the training of the hierarchical clustering algorithm based at least in part on the first factor, the second factor, and the third factor.

The invention relates to the technical field of major activity medical security communication, and discloses a major activity medical security information system based on multi-terminal positioning, and the system comprises the steps: obtaining a to-be-reported medical event data packet of a medical terminal, and generating a concurrent event feature set; calculating a comprehensive correlation degree value between every two events, and generating an event correlation degree matrix; clustering by using a hierarchical clustering algorithm to generate an event cluster division result; acquiring position tracking data of the movable large-scale equipment, and generating a dynamic blind area prediction space-time sequence; performing space-time cross judgment, and generating an event cluster communication vulnerability window; searching alternative relay terminals, and generating a cluster-level cooperative communication path reservation scheme; synchronously issuing a communication path switching instruction to all terminals in the event cluster at the arrival time of the communication vulnerability window; according to the invention, the problem of disordered reporting time sequence of the associated medical event information in a dynamic shielding scene is solved.

The invention relates to the technical field of oil well fracturing prediction, and solves the technical problems in the prior art that distribution of important features in fracturing data cannot be fully reserved generally through simple data segmentation, and low result accuracy in fracturing state prediction is easily caused. In particular to a fracturing state prediction system and method applying a machine learning intelligent decision-making system, and the method comprises the following steps: S1, obtaining original data of a gulongshale oil fracturing detection system, preprocessing the original data, and obtaining a unified table through an SQLJOIN method; the statistical feature extraction algorithm and the hierarchical clustering algorithm are used to construct a more excellent algorithm model, the accuracy and robustness of feature expression are improved by realizing feature extraction and clustering, the feature redundancy is reduced by combining statistical features and hierarchical clustering, the interpretability and prediction precision of a subsequent model are improved, and the prediction efficiency is improved. And meanwhile, the nonlinear relationship in the gulonium shale oil data is optimized, so that the subsequent steps are more accurate and smoother.

The invention discloses a transaction account state determination method and device, electronic equipment and a storage medium, and relates to the field of financial science and technology. The method comprises the following steps: acquiring a feature vector of each account in N accounts; all the feature vectors corresponding to the N accounts are clustered based on a preset first radius threshold value and a preset second radius threshold value, M feature subsets are obtained, a global clustering structure chart is constructed according to the M feature subsets, and the global clustering structure chart is used for clustering the N accounts according to the local hierarchical relation reflected by the M feature subsets. Representing fund flow paths in the N accounts in a tree topology form; and determining the account state of each account in the N accounts according to the global clustering structure diagram. The technical problem that in the prior art, when a hierarchical clustering algorithm is directly applied to full-amount bank card transaction data, calculation complexity is high, calculation efficiency and clustering precision cannot be considered at the same time, and abnormal account detection lags behind is solved.

The invention provides a micro-seismic monitoring and early warning method based on data purification and space-time weighted Markov clustering, and the method comprises the steps: carrying out the spatial boundary screening and statistical outlier elimination processing of original micro-seismic monitoring data, and obtaining an effective micro-seismic data set; key features are extracted from the effective micro-seismic data, a spatial component and a time component are separated, and statistics of the spatial component and the time component are calculated respectively; calculating a space-time weighted mahalanobis distance by combining space and time variability and correlation, and constructing a dissimilarity matrix between microseismic events based on the space-time weighted mahalanobis distance; carrying out clustering analysis by adopting a Ward hierarchical clustering algorithm based on the dissimilarity degree matrix, and dividing the microseismic events into different clusters; and mapping a clustering result back to an original data space, and generating and outputting various visual charts. According to the method, fine cleaning and high-precision spatial-temporal clustering of the micro-seismic data can be realized, and the data quality and rule recognition capability of mine micro-seismic monitoring can be remarkably improved.

The invention relates to a protein family automatic clustering method, system and device and a storage medium. The method comprises the following steps: inputting a to-be-clustered protein sequence into a protein language model for vector embedding to generate an embedded vector of the protein sequence; calculating the distance between every two protein sequences based on the embedding vectors of the protein sequences to obtain the pairwise distance between every two protein sequences, and measuring the similarity between the protein sequences according to the pairwise distance; carrying out bottom-up hierarchical clustering on the embedded vector of the protein sequence by adopting a hierarchical clustering algorithm, and constructing a tree structure of the protein sequence; based on the tree structure of the protein sequence, layer-by-layer clustering division is carried out according to set clustering parameters, classes meeting the clustering parameters are divided into the same protein family, and clustering division is ended until a clustering termination condition is met. According to the method, the distant homologous recognition capability is enhanced, and division deviation caused by the fact that fixed clustering parameters are difficult to adapt to different protein families at the same time and the difference degree in each family is inconsistent is avoided.

The invention discloses a compound eye camera image recognition method and system based on ensemble learning, and the method comprises the following steps: S1, constructing a base learner for each sub-eye of a compound eye camera, selecting an LSTM network as a learning algorithm of the base learner, taking the pixel coordinates of a key point collected by the sub-eye as the input of the LSTM network, and carrying out the learning of the LSTM network; taking the three-dimensional coordinates of the corresponding images as the output of the LSTM network, and establishing a mapping relation between the pixel coordinates of the remote sensing images collected by the compound eye camera and the three-dimensional coordinates; s2, clustering the output results of the base learners through a hierarchical clustering algorithm, selecting the output results of the base learners with errors smaller than a first threshold value, and rejecting the base learners with errors larger than a second threshold value; the second threshold is greater than the first threshold; and S3, weighted integration is carried out, so that the prediction precision of the integrated model is improved, and the calculation complexity of the integrated model can be reduced. Based on the idea of integrated learning, three-dimensional measurement and target positioning of the compound eye camera are realized, and the measurement precision can be effectively improved.

The application discloses a railway scene three-dimensional laser point cloud data denoising method and device, and relates to the field of railway mobile measurement technology, and the method comprises the steps of: acquiring point cloud data along a railway through a laser point cloud acquisition device, and dividing the point cloud data into multiple point cloud data segments; acquiring multiple coordinate values obtained by projecting each point cloud data segment onto a two-dimensional plane; dividing the two-dimensional plane into multiple grids, and determining multiple grid index points corresponding to the multiple coordinate values; taking each grid index point vector as a clustering center, respectively clustering the multiple grid index points, and determining multiple initial categories of the grid index points; according to a hierarchical clustering algorithm, merging each two initial categories in the multiple initial categories of the grid index points, obtaining multiple final categories of the grid index points and the number of grid index points contained in each final category; determining noise points in the multiple grid index points, determining and removing noise data corresponding to the noise points in the point cloud data segment, and quickly and efficiently denoising.